pg3066

File Attachment

Thumbnailjpg

ADVANCED WIRELESSNETWORKS

ADVANCED WIRELESSNETWORKSTECHNOLOGY AND BUSINESSMODELS

Third Edition

Savo GlisicUniversity of Oulu Finland

This edition first published 2016copy 2016 John Wiley amp Sons Ltd

First Edition published in 2006

Registered OfficeJohn Wiley amp Sons Ltd The Atrium Southern Gate Chichester West Sussex PO19 8SQ United Kingdom

For details of our global editorial offices for customer services and for information about how to apply forpermission to reuse the copyright material in this book please see our website at wwwwileycom

The right of the author to be identified as the author of this work has been asserted in accordance with theCopyright Designs and Patents Act 1988

All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted inany form or by any means electronic mechanical photocopying recording or otherwise except as permitted by theUK Copyright Designs and Patents Act 1988 without the prior permission of the publisher

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not beavailable in electronic books

Designations used by companies to distinguish their products are often claimed as trademarks All brand namesand product names used in this book are trade names service marks trademarks or registered trademarks of theirrespective owners The publisher is not associated with any product or vendor mentioned in this book

Limit of LiabilityDisclaimer of Warranty While the publisher and author have used their best efforts in preparing thisbook they make no representations or warranties with respect to the accuracy or completeness of the contents ofthis book and specifically disclaim any impliedwarranties of merchantability or fitness for a particular purpose It is soldon the understanding that the publisher is not engaged in rendering professional services and neither the publishernor the author shall be liable for damages arising herefrom If professional advice or other expert assistance is requiredthe services of a competent professional should be sought

Library of Congress Cataloging-in-Publication data applied for

ISBN 9781119096856

A catalogue record for this book is available from the British Library

Set in 10 12pt Times by SPi Global Pondicherry India

1 2016

ADVANCED WIRELESSNETWORKS

ADVANCED WIRELESSNETWORKSTECHNOLOGY AND BUSINESSMODELS

Third Edition

Savo GlisicUniversity of Oulu Finland

This edition first published 2016copy 2016 John Wiley amp Sons Ltd

First Edition published in 2006

Registered OfficeJohn Wiley amp Sons Ltd The Atrium Southern Gate Chichester West Sussex PO19 8SQ United Kingdom

For details of our global editorial offices for customer services and for information about how to apply forpermission to reuse the copyright material in this book please see our website at wwwwileycom

The right of the author to be identified as the author of this work has been asserted in accordance with theCopyright Designs and Patents Act 1988

All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted inany form or by any means electronic mechanical photocopying recording or otherwise except as permitted by theUK Copyright Designs and Patents Act 1988 without the prior permission of the publisher

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not beavailable in electronic books

Designations used by companies to distinguish their products are often claimed as trademarks All brand namesand product names used in this book are trade names service marks trademarks or registered trademarks of theirrespective owners The publisher is not associated with any product or vendor mentioned in this book

Limit of LiabilityDisclaimer of Warranty While the publisher and author have used their best efforts in preparing thisbook they make no representations or warranties with respect to the accuracy or completeness of the contents ofthis book and specifically disclaim any impliedwarranties of merchantability or fitness for a particular purpose It is soldon the understanding that the publisher is not engaged in rendering professional services and neither the publishernor the author shall be liable for damages arising herefrom If professional advice or other expert assistance is requiredthe services of a competent professional should be sought

Library of Congress Cataloging-in-Publication data applied for

ISBN 9781119096856

A catalogue record for this book is available from the British Library

Set in 10 12pt Times by SPi Global Pondicherry India

1 2016

ADVANCED WIRELESSNETWORKSTECHNOLOGY AND BUSINESSMODELS

Third Edition

Savo GlisicUniversity of Oulu Finland

This edition first published 2016copy 2016 John Wiley amp Sons Ltd

First Edition published in 2006

Registered OfficeJohn Wiley amp Sons Ltd The Atrium Southern Gate Chichester West Sussex PO19 8SQ United Kingdom

For details of our global editorial offices for customer services and for information about how to apply forpermission to reuse the copyright material in this book please see our website at wwwwileycom

The right of the author to be identified as the author of this work has been asserted in accordance with theCopyright Designs and Patents Act 1988

All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted inany form or by any means electronic mechanical photocopying recording or otherwise except as permitted by theUK Copyright Designs and Patents Act 1988 without the prior permission of the publisher

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not beavailable in electronic books

Designations used by companies to distinguish their products are often claimed as trademarks All brand namesand product names used in this book are trade names service marks trademarks or registered trademarks of theirrespective owners The publisher is not associated with any product or vendor mentioned in this book

Limit of LiabilityDisclaimer of Warranty While the publisher and author have used their best efforts in preparing thisbook they make no representations or warranties with respect to the accuracy or completeness of the contents ofthis book and specifically disclaim any impliedwarranties of merchantability or fitness for a particular purpose It is soldon the understanding that the publisher is not engaged in rendering professional services and neither the publishernor the author shall be liable for damages arising herefrom If professional advice or other expert assistance is requiredthe services of a competent professional should be sought

Library of Congress Cataloging-in-Publication data applied for

ISBN 9781119096856

A catalogue record for this book is available from the British Library

Set in 10 12pt Times by SPi Global Pondicherry India

1 2016

This edition first published 2016copy 2016 John Wiley amp Sons Ltd

First Edition published in 2006

Registered OfficeJohn Wiley amp Sons Ltd The Atrium Southern Gate Chichester West Sussex PO19 8SQ United Kingdom

For details of our global editorial offices for customer services and for information about how to apply forpermission to reuse the copyright material in this book please see our website at wwwwileycom

The right of the author to be identified as the author of this work has been asserted in accordance with theCopyright Designs and Patents Act 1988

All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted inany form or by any means electronic mechanical photocopying recording or otherwise except as permitted by theUK Copyright Designs and Patents Act 1988 without the prior permission of the publisher

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not beavailable in electronic books

Designations used by companies to distinguish their products are often claimed as trademarks All brand namesand product names used in this book are trade names service marks trademarks or registered trademarks of theirrespective owners The publisher is not associated with any product or vendor mentioned in this book

Limit of LiabilityDisclaimer of Warranty While the publisher and author have used their best efforts in preparing thisbook they make no representations or warranties with respect to the accuracy or completeness of the contents ofthis book and specifically disclaim any impliedwarranties of merchantability or fitness for a particular purpose It is soldon the understanding that the publisher is not engaged in rendering professional services and neither the publishernor the author shall be liable for damages arising herefrom If professional advice or other expert assistance is requiredthe services of a competent professional should be sought

Library of Congress Cataloging-in-Publication data applied for

ISBN 9781119096856

A catalogue record for this book is available from the British Library

Set in 10 12pt Times by SPi Global Pondicherry India

1 2016

Contents

Preface xv

1 Introduction Generalized Model of Advanced Wireless Networks 111 Network Model 3

111 Node Percolation 3112 Link PercolationmdashCognitive Links 4

12 Network Connectivity 513 Wireless Network Design with Small World Properties 7

131 Cell Rewiring 7132 Traffic Distribution Aware Rewiring 9133 Multicell Rewiring 10

14 Frequency Channels Backup 11141 mkfs Contract 11142 Random Redundancy Assignment (R2A) 11143 On Demand Redundancy Assignment 12

15 Generalized Network Model 1316 Routing Protocols Over s-Lattice Network 14

161 Application Specific Routing Protocol 1617 Network Performance 16

171 Average Path Length 18172 Clustering 18

18 Node Route Topology and Network Robustness 1919 Power Consumption 20110 Protocol Complexity 20111 Performance Evaluation 21

1111 Average Path Length 211112 Clustering 231113 Node Robustness 231114 Network Robustness 23

1115 Power Consumption 241116 Protocol Complexity 25

112 Book Layout 271121 Chapter 1 Introduction Generalized Model of Advanced

Wireless Networks 281122 Chapter 2 Adaptive Network Layer 281123 Chapter 3 Mobility Management 281124 Chapter 4 Ad Hoc Networks 281125 Chapter 5 Sensor Networks 281126 Chapter 6 Security 291127 Chapter 7 Networks Economy 291128 Chapter 8 Multi-Hop Cellular Networks 291129 Chapter 9 Cognitive Networks 2911210 Chapter 10 Stochastic Geometry 2911211 Chapter 11 Heterogeneous Networks 3011212 Chapter 12 Access Point Selection 3011213 Chapter 13 Self-Organizing Networks 3011214 Chapter 14 Complex Networks 3011215 Chapter 15 Massive MIMO 3011216 Chapter 16 Network Optimization Theory 3111217 Chapter 17 Network Information Theory 3111218 Chapter 18 Network Stability 3111219 Chapter 19 Multi-Operator Spectrum Sharing 3111220 Chapter 20 Large Scale Networks and Mean Field Theory 3111221 Chapter 21 mmWave 3D Networks 3211222 Chapter 22 Cloud Computing in Wireless Network 3211223 Chapter 23 Wireless Networks and Matching Theory 3211224 Chapter 24 Dynamic Wireless Network Infrastructure 33

Appendix A1 33References 34

2 Adaptive Network Layer 3521 Graphs and Routing Protocols 3522 Graph Theory 5423 Routing with Topology Aggregation 56

231 Network and Aggregation Models 58References 60

3 Mobility Management 6531 Cellular Networks 65

311 Mobility Management in Cellular Networks 67312 Location Registration and Call Delivery 71313 Location Update and Terminal Paging 74314 WATM Handoff Management in 4G Wireless Networks 88315 Mobility Management for Satellite Networks 89

vi Contents

32 Cellular Systems with Prioritized Handoff 89321 Performance Examples 99

33 Cell Residing Time Distribution 10034 Mobility Prediction in Pico- and Micro-Cellular Networks 105

341 PST-QoS Guarantees Framework 107342 Most Likely Cluster Model 108

Appendix A3 Distance Calculation in an Intermediate Cell 116References 122

4 Ad Hoc Networks 12641 Routing Protocols 126

411 Ad Hoc Routing Protocols 127412 Reactive Protocols 134

42 Hybrid Routing Protocol 14643 Scalable Routing Strategies 15244 Multipath Routing 16045 Clustering Protocols 162

451 Introduction 162452 Clustering Algorithm 164

46 Cashing Schemes for Routing 17547 Distributed QoS Routing 181

471 Forwarding the Received Tokens 185472 Bandwidth Constrained Routing 186473 Forwarding the Received Tokens 187

References 190

5 Sensor Networks 19451 Introduction 19452 Sensor Network Parameters 19653 Sensor Network Architecture 199

531 Physical Layer 199532 Data Link Layer 200533 Network Layer 202534 Transport Layer 207535 Application Layer 208

54 Mobile Sensor Network Deployment 20955 Directed Diffusion 21256 Aggregation in Wireless Sensor Networks 21657 Boundary Estimation 220

571 Number of RDPs in 222572 Kraft Inequality 222573 Upper Bounds on Achievable Accuracy 223574 System Optimization 224

58 Optimal Transmission Radius in Sensor Networks 22759 Data Funneling 233510 Equivalent Transport Control Protocol in Sensor Networks 236References 237

viiContents

6 Security 24461 Authentication 244

611 Attacks on Simple Cryptographic Authentication 247612 Canonical Authentication Protocol 250

62 Security Architecture 25363 Key Management 25764 Security in Ad Hoc Networks 261

641 Self-Organized Key Management 26565 Security in Sensor Networks 268References 269

7 Network Economics 27271 Fundamentals of Network Economics 272

711 Externalities 273712 Pricing of Services 274713 Congestion Pricing 275714 Congestion Game 276715 Modeling Service Differentiation 277716 Competition 278717 Auctions 279718 Bidding for QoS 280719 Bandwidth Auction 2817110 Investments 282

72 Wireless Network Microeconomics Data Sponsoring 286721 Background Solutions 287722 Sponsored Data Model 287

73 Spectrum Pricing for Market Equilibrium 291731 Network and Pricing Model 291732 Optimization of Spectrum Pricing 292733 Distributed Solutions 295734 Stability of Distributed Pricing Models 297

74 Sequential Spectrum Sharing 300741 Sequential Spectrum Sharing and Interrelated Market Model 301742 Iterative Negotiation Algorithms 304

75 Data Plan Trading 308751 Modeling Userrsquos BuyerSeller Trading Incentives 309752 ISP Trading Policies 313

References 315

8 Multi-Hop Cellular Networks 31881 Modeling Multi-Hop Multi-Operator Multi-Technology Wireless Networks 31882 Technology Background 31983 System Model and Notation 32184 m3 Route Discovery Protocols 323

841 Minimum Distance Routing 323842 Limited Interference RoutingScheduling 324

viii Contents

85 Performance of m3 Route Discovery Protocols 32786 Protocol Complexity 32987 Traffic Offloading Incentives 330

871 Collaborative Negotiation between MNO and SSO 33588 Performance Illustrations 335

881 m3 Route Discovery Protocols 336882 Capacity and Throughput for the Modified m3 Route

Discovery Protocols 338883 Traffic Offloading Incentives 341884 Implementation and Impact of Mobility 343

References 344

9 Cognitive Networks 34691 Technology Background 346

911 Fundamentals 346912 Network and Transport Layer Protocols 348

92 Spectrum Auctions for Multi-hop Cognitive Networks 350921 Background Technology 352922 System Model 353923 Heuristic Truthful Auction 356924 Randomized Auction 359

93 Compound Auctioning in Multi-hop Cognitive Cellular Networks 363931 Network Model 364932 Spectrum Aware Routing Discovery Protocol 367933 Joint Resource Auction and Tipping Scheme 370934 Reinforcement Learning Based Auction Scheme 372935 Group Buying Based Auction Design 373936 Further Extension to General Scenarios 377937 System Performance 378

References 388

10 Stochastic Geometry 391101 Background Theory 391

1011 Point Process 3911012 Outage Probability 3941013 Multi-tier Networks 396

References 398

11 Heterogeneous Networks 402111 Preliminaries 402112 Self-Organized Small Cell Networks 404

1121 Background Technology 4041122 System Model 4051123 Self-Organized SCN 409

113 Dynamic Network Architecture 4111131 System Model 412

ixContents

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

1115 Power Consumption 241116 Protocol Complexity 25

112 Book Layout 271121 Chapter 1 Introduction Generalized Model of Advanced

Wireless Networks 281122 Chapter 2 Adaptive Network Layer 281123 Chapter 3 Mobility Management 281124 Chapter 4 Ad Hoc Networks 281125 Chapter 5 Sensor Networks 281126 Chapter 6 Security 291127 Chapter 7 Networks Economy 291128 Chapter 8 Multi-Hop Cellular Networks 291129 Chapter 9 Cognitive Networks 2911210 Chapter 10 Stochastic Geometry 2911211 Chapter 11 Heterogeneous Networks 3011212 Chapter 12 Access Point Selection 3011213 Chapter 13 Self-Organizing Networks 3011214 Chapter 14 Complex Networks 3011215 Chapter 15 Massive MIMO 3011216 Chapter 16 Network Optimization Theory 3111217 Chapter 17 Network Information Theory 3111218 Chapter 18 Network Stability 3111219 Chapter 19 Multi-Operator Spectrum Sharing 3111220 Chapter 20 Large Scale Networks and Mean Field Theory 3111221 Chapter 21 mmWave 3D Networks 3211222 Chapter 22 Cloud Computing in Wireless Network 3211223 Chapter 23 Wireless Networks and Matching Theory 3211224 Chapter 24 Dynamic Wireless Network Infrastructure 33

Appendix A1 33References 34

2 Adaptive Network Layer 3521 Graphs and Routing Protocols 3522 Graph Theory 5423 Routing with Topology Aggregation 56

231 Network and Aggregation Models 58References 60

3 Mobility Management 6531 Cellular Networks 65

311 Mobility Management in Cellular Networks 67312 Location Registration and Call Delivery 71313 Location Update and Terminal Paging 74314 WATM Handoff Management in 4G Wireless Networks 88315 Mobility Management for Satellite Networks 89

vi Contents

32 Cellular Systems with Prioritized Handoff 89321 Performance Examples 99

33 Cell Residing Time Distribution 10034 Mobility Prediction in Pico- and Micro-Cellular Networks 105

341 PST-QoS Guarantees Framework 107342 Most Likely Cluster Model 108

Appendix A3 Distance Calculation in an Intermediate Cell 116References 122

4 Ad Hoc Networks 12641 Routing Protocols 126

411 Ad Hoc Routing Protocols 127412 Reactive Protocols 134

42 Hybrid Routing Protocol 14643 Scalable Routing Strategies 15244 Multipath Routing 16045 Clustering Protocols 162

451 Introduction 162452 Clustering Algorithm 164

46 Cashing Schemes for Routing 17547 Distributed QoS Routing 181

471 Forwarding the Received Tokens 185472 Bandwidth Constrained Routing 186473 Forwarding the Received Tokens 187

References 190

5 Sensor Networks 19451 Introduction 19452 Sensor Network Parameters 19653 Sensor Network Architecture 199

531 Physical Layer 199532 Data Link Layer 200533 Network Layer 202534 Transport Layer 207535 Application Layer 208

54 Mobile Sensor Network Deployment 20955 Directed Diffusion 21256 Aggregation in Wireless Sensor Networks 21657 Boundary Estimation 220

571 Number of RDPs in 222572 Kraft Inequality 222573 Upper Bounds on Achievable Accuracy 223574 System Optimization 224

58 Optimal Transmission Radius in Sensor Networks 22759 Data Funneling 233510 Equivalent Transport Control Protocol in Sensor Networks 236References 237

viiContents

6 Security 24461 Authentication 244

611 Attacks on Simple Cryptographic Authentication 247612 Canonical Authentication Protocol 250

62 Security Architecture 25363 Key Management 25764 Security in Ad Hoc Networks 261

641 Self-Organized Key Management 26565 Security in Sensor Networks 268References 269

7 Network Economics 27271 Fundamentals of Network Economics 272

711 Externalities 273712 Pricing of Services 274713 Congestion Pricing 275714 Congestion Game 276715 Modeling Service Differentiation 277716 Competition 278717 Auctions 279718 Bidding for QoS 280719 Bandwidth Auction 2817110 Investments 282

72 Wireless Network Microeconomics Data Sponsoring 286721 Background Solutions 287722 Sponsored Data Model 287

73 Spectrum Pricing for Market Equilibrium 291731 Network and Pricing Model 291732 Optimization of Spectrum Pricing 292733 Distributed Solutions 295734 Stability of Distributed Pricing Models 297

74 Sequential Spectrum Sharing 300741 Sequential Spectrum Sharing and Interrelated Market Model 301742 Iterative Negotiation Algorithms 304

75 Data Plan Trading 308751 Modeling Userrsquos BuyerSeller Trading Incentives 309752 ISP Trading Policies 313

References 315

8 Multi-Hop Cellular Networks 31881 Modeling Multi-Hop Multi-Operator Multi-Technology Wireless Networks 31882 Technology Background 31983 System Model and Notation 32184 m3 Route Discovery Protocols 323

841 Minimum Distance Routing 323842 Limited Interference RoutingScheduling 324

viii Contents

85 Performance of m3 Route Discovery Protocols 32786 Protocol Complexity 32987 Traffic Offloading Incentives 330

871 Collaborative Negotiation between MNO and SSO 33588 Performance Illustrations 335

881 m3 Route Discovery Protocols 336882 Capacity and Throughput for the Modified m3 Route

Discovery Protocols 338883 Traffic Offloading Incentives 341884 Implementation and Impact of Mobility 343

References 344

9 Cognitive Networks 34691 Technology Background 346

911 Fundamentals 346912 Network and Transport Layer Protocols 348

92 Spectrum Auctions for Multi-hop Cognitive Networks 350921 Background Technology 352922 System Model 353923 Heuristic Truthful Auction 356924 Randomized Auction 359

93 Compound Auctioning in Multi-hop Cognitive Cellular Networks 363931 Network Model 364932 Spectrum Aware Routing Discovery Protocol 367933 Joint Resource Auction and Tipping Scheme 370934 Reinforcement Learning Based Auction Scheme 372935 Group Buying Based Auction Design 373936 Further Extension to General Scenarios 377937 System Performance 378

References 388

10 Stochastic Geometry 391101 Background Theory 391

1011 Point Process 3911012 Outage Probability 3941013 Multi-tier Networks 396

References 398

11 Heterogeneous Networks 402111 Preliminaries 402112 Self-Organized Small Cell Networks 404

1121 Background Technology 4041122 System Model 4051123 Self-Organized SCN 409

113 Dynamic Network Architecture 4111131 System Model 412

ixContents

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

32 Cellular Systems with Prioritized Handoff 89321 Performance Examples 99

33 Cell Residing Time Distribution 10034 Mobility Prediction in Pico- and Micro-Cellular Networks 105

341 PST-QoS Guarantees Framework 107342 Most Likely Cluster Model 108

Appendix A3 Distance Calculation in an Intermediate Cell 116References 122

4 Ad Hoc Networks 12641 Routing Protocols 126

411 Ad Hoc Routing Protocols 127412 Reactive Protocols 134

42 Hybrid Routing Protocol 14643 Scalable Routing Strategies 15244 Multipath Routing 16045 Clustering Protocols 162

451 Introduction 162452 Clustering Algorithm 164

46 Cashing Schemes for Routing 17547 Distributed QoS Routing 181

471 Forwarding the Received Tokens 185472 Bandwidth Constrained Routing 186473 Forwarding the Received Tokens 187

References 190

5 Sensor Networks 19451 Introduction 19452 Sensor Network Parameters 19653 Sensor Network Architecture 199

531 Physical Layer 199532 Data Link Layer 200533 Network Layer 202534 Transport Layer 207535 Application Layer 208

54 Mobile Sensor Network Deployment 20955 Directed Diffusion 21256 Aggregation in Wireless Sensor Networks 21657 Boundary Estimation 220

571 Number of RDPs in 222572 Kraft Inequality 222573 Upper Bounds on Achievable Accuracy 223574 System Optimization 224

58 Optimal Transmission Radius in Sensor Networks 22759 Data Funneling 233510 Equivalent Transport Control Protocol in Sensor Networks 236References 237

viiContents

6 Security 24461 Authentication 244

611 Attacks on Simple Cryptographic Authentication 247612 Canonical Authentication Protocol 250

62 Security Architecture 25363 Key Management 25764 Security in Ad Hoc Networks 261

641 Self-Organized Key Management 26565 Security in Sensor Networks 268References 269

7 Network Economics 27271 Fundamentals of Network Economics 272

711 Externalities 273712 Pricing of Services 274713 Congestion Pricing 275714 Congestion Game 276715 Modeling Service Differentiation 277716 Competition 278717 Auctions 279718 Bidding for QoS 280719 Bandwidth Auction 2817110 Investments 282

72 Wireless Network Microeconomics Data Sponsoring 286721 Background Solutions 287722 Sponsored Data Model 287

73 Spectrum Pricing for Market Equilibrium 291731 Network and Pricing Model 291732 Optimization of Spectrum Pricing 292733 Distributed Solutions 295734 Stability of Distributed Pricing Models 297

74 Sequential Spectrum Sharing 300741 Sequential Spectrum Sharing and Interrelated Market Model 301742 Iterative Negotiation Algorithms 304

75 Data Plan Trading 308751 Modeling Userrsquos BuyerSeller Trading Incentives 309752 ISP Trading Policies 313

References 315

8 Multi-Hop Cellular Networks 31881 Modeling Multi-Hop Multi-Operator Multi-Technology Wireless Networks 31882 Technology Background 31983 System Model and Notation 32184 m3 Route Discovery Protocols 323

841 Minimum Distance Routing 323842 Limited Interference RoutingScheduling 324

viii Contents

85 Performance of m3 Route Discovery Protocols 32786 Protocol Complexity 32987 Traffic Offloading Incentives 330

871 Collaborative Negotiation between MNO and SSO 33588 Performance Illustrations 335

881 m3 Route Discovery Protocols 336882 Capacity and Throughput for the Modified m3 Route

Discovery Protocols 338883 Traffic Offloading Incentives 341884 Implementation and Impact of Mobility 343

References 344

9 Cognitive Networks 34691 Technology Background 346

911 Fundamentals 346912 Network and Transport Layer Protocols 348

92 Spectrum Auctions for Multi-hop Cognitive Networks 350921 Background Technology 352922 System Model 353923 Heuristic Truthful Auction 356924 Randomized Auction 359

93 Compound Auctioning in Multi-hop Cognitive Cellular Networks 363931 Network Model 364932 Spectrum Aware Routing Discovery Protocol 367933 Joint Resource Auction and Tipping Scheme 370934 Reinforcement Learning Based Auction Scheme 372935 Group Buying Based Auction Design 373936 Further Extension to General Scenarios 377937 System Performance 378

References 388

10 Stochastic Geometry 391101 Background Theory 391

1011 Point Process 3911012 Outage Probability 3941013 Multi-tier Networks 396

References 398

11 Heterogeneous Networks 402111 Preliminaries 402112 Self-Organized Small Cell Networks 404

1121 Background Technology 4041122 System Model 4051123 Self-Organized SCN 409

113 Dynamic Network Architecture 4111131 System Model 412

ixContents

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

6 Security 24461 Authentication 244

611 Attacks on Simple Cryptographic Authentication 247612 Canonical Authentication Protocol 250

62 Security Architecture 25363 Key Management 25764 Security in Ad Hoc Networks 261

641 Self-Organized Key Management 26565 Security in Sensor Networks 268References 269

7 Network Economics 27271 Fundamentals of Network Economics 272

711 Externalities 273712 Pricing of Services 274713 Congestion Pricing 275714 Congestion Game 276715 Modeling Service Differentiation 277716 Competition 278717 Auctions 279718 Bidding for QoS 280719 Bandwidth Auction 2817110 Investments 282

72 Wireless Network Microeconomics Data Sponsoring 286721 Background Solutions 287722 Sponsored Data Model 287

73 Spectrum Pricing for Market Equilibrium 291731 Network and Pricing Model 291732 Optimization of Spectrum Pricing 292733 Distributed Solutions 295734 Stability of Distributed Pricing Models 297

74 Sequential Spectrum Sharing 300741 Sequential Spectrum Sharing and Interrelated Market Model 301742 Iterative Negotiation Algorithms 304

75 Data Plan Trading 308751 Modeling Userrsquos BuyerSeller Trading Incentives 309752 ISP Trading Policies 313

References 315

8 Multi-Hop Cellular Networks 31881 Modeling Multi-Hop Multi-Operator Multi-Technology Wireless Networks 31882 Technology Background 31983 System Model and Notation 32184 m3 Route Discovery Protocols 323

841 Minimum Distance Routing 323842 Limited Interference RoutingScheduling 324

viii Contents

85 Performance of m3 Route Discovery Protocols 32786 Protocol Complexity 32987 Traffic Offloading Incentives 330

871 Collaborative Negotiation between MNO and SSO 33588 Performance Illustrations 335

881 m3 Route Discovery Protocols 336882 Capacity and Throughput for the Modified m3 Route

Discovery Protocols 338883 Traffic Offloading Incentives 341884 Implementation and Impact of Mobility 343

References 344

9 Cognitive Networks 34691 Technology Background 346

911 Fundamentals 346912 Network and Transport Layer Protocols 348

92 Spectrum Auctions for Multi-hop Cognitive Networks 350921 Background Technology 352922 System Model 353923 Heuristic Truthful Auction 356924 Randomized Auction 359

93 Compound Auctioning in Multi-hop Cognitive Cellular Networks 363931 Network Model 364932 Spectrum Aware Routing Discovery Protocol 367933 Joint Resource Auction and Tipping Scheme 370934 Reinforcement Learning Based Auction Scheme 372935 Group Buying Based Auction Design 373936 Further Extension to General Scenarios 377937 System Performance 378

References 388

10 Stochastic Geometry 391101 Background Theory 391

1011 Point Process 3911012 Outage Probability 3941013 Multi-tier Networks 396

References 398

11 Heterogeneous Networks 402111 Preliminaries 402112 Self-Organized Small Cell Networks 404

1121 Background Technology 4041122 System Model 4051123 Self-Organized SCN 409

113 Dynamic Network Architecture 4111131 System Model 412

ixContents

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

85 Performance of m3 Route Discovery Protocols 32786 Protocol Complexity 32987 Traffic Offloading Incentives 330

871 Collaborative Negotiation between MNO and SSO 33588 Performance Illustrations 335

881 m3 Route Discovery Protocols 336882 Capacity and Throughput for the Modified m3 Route

Discovery Protocols 338883 Traffic Offloading Incentives 341884 Implementation and Impact of Mobility 343

References 344

9 Cognitive Networks 34691 Technology Background 346

911 Fundamentals 346912 Network and Transport Layer Protocols 348

92 Spectrum Auctions for Multi-hop Cognitive Networks 350921 Background Technology 352922 System Model 353923 Heuristic Truthful Auction 356924 Randomized Auction 359

93 Compound Auctioning in Multi-hop Cognitive Cellular Networks 363931 Network Model 364932 Spectrum Aware Routing Discovery Protocol 367933 Joint Resource Auction and Tipping Scheme 370934 Reinforcement Learning Based Auction Scheme 372935 Group Buying Based Auction Design 373936 Further Extension to General Scenarios 377937 System Performance 378

References 388

10 Stochastic Geometry 391101 Background Theory 391

1011 Point Process 3911012 Outage Probability 3941013 Multi-tier Networks 396

References 398

11 Heterogeneous Networks 402111 Preliminaries 402112 Self-Organized Small Cell Networks 404

1121 Background Technology 4041122 System Model 4051123 Self-Organized SCN 409

113 Dynamic Network Architecture 4111131 System Model 412

ixContents

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

1132 Optimum Network Architecture 4151133 Dynamic Tracking of the Optimum Topology 4221134 Performance Illustrations 427

114 Economics of Heterogeneous Networks 4341141 Macrocell Service Only 4341142 Introducing Femtocells 4361143 Impact of Usersrsquo Reservation Payoffs 4381144 Femtocell Frequency Reuse 4401145 Femtocell Operational Cost 4401146 Limited Femtocell Coverage 441

References 443

12 Access Point Selection 446121 Background Technology 446122 Network Selection Game 449123 Joint Access Point Selection and Power Allocation 453

1231 Single AP Network 4541232 Joint AP Selection and Power Control 4571233 Distributed Algorithms 459

124 Joint AP Selection and Beamforming Optimization 4631241 Network Model 463

References 474

13 Self-Organizing Networks 478131 Self-Organizing Network Optimization 478132 System Model 478133 Joint Optimization of Tilts and AP Association 481

1331 System Objective Function 4811332 Optimization Problem 482

References 484

14 Complex Networks 486141 Evolution Towards Large-Scale Networks 486

1411 Types of Networks 487142 Network Characteristics 491143 Random Graphs 494References 496

15 Massive MIMO 499151 Linearly Precoded Multicellular Downlink System 499

1511 Background Technology 500152 System Model 503

1521 Channel Uncertainty Modeling 5041522 Stochastic Optimization 505

153 Optimization for Perfect Channel State Information 505

x Contents

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

154 Robust Designs for WSRM Problem 5091541 Approximation 1 5101542 Approximation 2 512

Appendix A15 519Appendix B15 519References 521

16 Network Optimization Theory 523161 Introduction 523162 Layering as Optimization Decomposition 524

1621 TCP Congestion Control 5251622 TCP RenoRED 5261623 TCP VegasDropTail 5261624 Optimization of MAC Protocol 5271625 Utility Optimal MAC ProtocolSocial Optimum 530

163 Cross-Layer Optimization 5331631 Congestion Control and Routing 5331632 Congestion Control and Physical Resource Allocation 5361633 Congestion and Contention Control 5381634 Congestion Control Routing and Scheduling 542

164 Optimization Problem Decomposition Methods 5431641 Decoupling Coupled Constraints 5431642 Dual Decomposition of the Basic NUM 5441643 Coupling Constraints 5471644 Decoupling Coupled Objectives 5481645 Alternative Decompositions 550

References 554

17 Network Information Theory 557171 Capacity of Ad Hoc Networks 557

1711 Arbitrary Networks 5581712 Random Networks 5591713 Arbitrary Networks Upper Bound on Transport Capacity 5601714 Arbitrary Networks Lower Bound on Transport Capacity 5641715 Random Networks Lower Bound on Throughput Capacity 565

172 Information Theory and Network Architectures 5691721 Upper Bounds Under High Attenuation 5711722 Multihop and Feasible Lower Bounds Under High Attenuation 573

173 Cooperative Transmission in Wireless Multihop Ad Hoc Networks 577References 584

18 Stability of Advanced Network Architectures 585181 Stability of Cooperative Cognitive Wireless Networks 585182 System Model 586

1821 Network Architecture 5861822 Channel 5881823 Cooperative Communication 589

xiContents

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

183 System Optimization 590184 Optimal Control Policy 592185 Achievable Rates 594

1851 Cooperative Network Stability Region 5941852 Non-Cooperative Network Stability Region 597

186 Stabilizing Transmission Policies 5981861 Network Parameters 5981862 Stationary Transmission Policy 5991863 Lyapynov Drift Analysis of the STAT Policy 6001864 Stability of the Dynamic Transmission Policy 604

References 605

19 Multi-Operator Spectrum Sharing 607191 Business Models for Spectrum Sharing 607

1911 Background Technology 6071912 Multi-Operator Cooperation Models 6101913 System Performance 6191914 Performance Illustrations 631

192 Spectrum Sharing in Multi-hop Networks 6381921 Multi-Operator Cooperation Models 6391922 System Analysis 6421923 System Performance 652

References 656

20 Large Scale Networks and Mean Field Theory 659201 MFT for Large Heterogeneous Cellular Networks 659

2011 System Model 6602012 Macro BS Optimization Problem 6602013 Mean Field Game Among Femto BSs 6622014 Interference Average Estimation 663

202 Large Scale Network Model Compression 6642021 Model Definition 6652022 Mean Field Analysis 667

203 Mean Field Theory Model of Large Scale DTN Networks 668204 Mean Field Modeling of Adaptive Infection Recovery in Multicast

DTN Networks 6742041 Background Technology 6742042 System Model 6772043 Recovery Schemes for Multicast DTN 6792044 System Performance 6842045 Extensions of the Model and Implementation Issues 6872046 Illustrations 690

205 Mean Field Theory for Scale-Free Random Networks 7012051 Network Models 7012052 The Scale-Free Model by Barabasi 703

xii Contents

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

2053 Mean Field Network Model 7052054 Incomplete BA Network Models 706

206 Spectrum Sharing and MFT 7092061 Optimal Wireless Service Provider Selection Strategy

Using MFT 7092062 WSP Selection Strategy for Finite Number of Terminals 711

207 Modeling Dynamics of Complex System 7112071 Dynamic System Model 7122072 BirthndashDeath Network Model 7172073 Network Rewiring 7192074 Multiple Time Scale System Optimization 719

Appendix A20 Iterative Algorithm to Solve Systems of Nonlinear ODEs(DiNSE-Algorithm) 721Appendix B20 Infection Rate of Destinations for DNCM 722Appendix C20 Infection Rate for Basic Epidemic Routing 722References 722

21 mmWave Networks 726211 mmWave Technology in Subcellular Architecture 726

2111 Limitations of mmWave Technology 7272112 Network Model 7282113 Network Performance 7312114 Performance of Dense mmWave Networks 735

212 Microeconomics of Dynamic mmWave Networks 7372121 Dynamic Small Cell Networks 7372122 DSC Network Model 7382123 DSC Network Performance 739

References 747

22 Cloud Computing in Wireless Networks 750221 Technology Background 750222 System Model 752223 System Optimization 756224 Dynamic Control Algorithm 758

2241 Resource Allocation at the Terminals 7582242 Resource Allocation at the Servers 761

225 Achievable Rates 7612251 Supportable Input Rate Region at the Terminals 7612252 Supportable Input Rate Region at Servers 763

226 Stabilizing Control Policies 7632261 Lyapunov Drift 7632262 Randomized Stationary Policy 7652263 Frame Based Policy 7662264 Dynamic Control Policy 767

References 769

xiiiContents

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

23 Wireless Networks and Matching Theory 771231 Background Technology Matching Markets 772

2311 Two-Sided Matching 7722312 One-Sided Matching 775

232 Distributed Stable Matching in Multiple Operator Cellular Network withTraffic Offloading 7762321 System Model 7772322 Problem Formulation 778

233 College Admissions Game Model for Cellular Networks withTraffic Offloading 7792331 System Model 7792332 Modeling Access Point Selection as College Admissions Matching 781

234 Many to Many Matching Games for Caching in Wireless Networks 7832341 System Model 7832342 Proactive Caching and Matching Theory 7842343 Proactive Caching Algorithm 786

235 Many to One Matching with Externalities in Cellular Networks withTraffic Offloading 7872351 System Model 7872352 Offloading Cell Association as a Matching Game with Externalities 789

236 Security in Matching of Device to Device Pairs in Cellular Networks 7912361 System Model 7922362 True Preferences 7932363 Cheating Coalition Strategy 794

References 795

24 Dynamic Wireless Network Infrastructure 797241 Infrastructure Sharing in Multi-Operator Cellular Networks 797

2411 System Model 7982412 Infrastructure Sharing Mechanism 799

242 User Provided Connectivity 8022421 System Model 8022422 Aggregate Service Value 804

243 Network Virtualization 8062431 Service-Oriented Network Virtualization in Telecommunications 807

244 Software Defined Networks 8102441 Current SDN Architectures 8112442 SDN Architecture Components 8122443 SDN Control Models 8132444 SDN and Infrastructure Based Wireless Access Networks 814

245 SDN Security 8162451 Security in Programmable Networks 8162452 Security Threats in SDN Networks 8172453 Security Solutions for SDN Network 818

References 819

Index 827

xiv Contents

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

Preface

Wireless communications has been developed so far through generations 1G to 4G withexclusive focus on improving the physical layer This concept has at least two drawbacks firstwireless channels cannot compete with optical networks when it comes to network capacitysecond the advantages of user mobility have not been emphasized enough In the scenarios offuture dense networks with a significant increase of user terminals and access points wirelesslinks in the wireless access concept in 5G will become shorter and shorter asking for morefrequent handoffs which jeopardize the reliability of the connectionsA significant part of the future networks will handle Internet of Things and People (IoTP)

communications where sophisticated physical layer solutions cannot be used Human bodyimplants will use simple solutions For these reasons there is a common understanding that5G will be about wireless networks rather than about wireless access to the networks In theresearch of the enabling technologies for 5G different communities focus on different solu-tions Small cell technology mmWave physical layer cognitive networks massive MIMOspectra and infrastructure sharing in multi-operator network management dynamic networkarchitecture user provided networks and so onIn thedesign andanalysis of thesenetworks anumber ofpowerful analytical tools are used like

convex dynamic and stochastic optimization stochastic geometry mean field theory matchingtheory and game theory as well as a number of tools used in economicsmicroeconomicsThis book advocates a concept where all these technologies will be simultaneously present in

the future wireless networks and focuses on three main issues

1 Design of heterogeneous networks that include all or a number of these technologies at thesame time

2 Optimization of such complex networks3 Design of efficient business models to exploit the limited resources of these networks

Hence the subtitle of this book Technology and Business ModelsThe book is dedicated to the young generation of open-minded researchers network design-

ers and managers who will make it happen

Savo Glisic

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

1IntroductionGeneralized Model of AdvancedWireless Networks

In the process of evolving towards 5G networks wireless networks are becoming more com-plex in both the number of different functionalities they provide as well as in the number ofusers they serve [1] Future 5G networks are expected to be highly heterogeneous (see Chapter11) and to integrate cognitive network concepts [2 3] (Chapter 9) heterogeneous solutions forthe offload of cellular network traffic toWLANs [4 5] multi-hop cellular networks (Chapter 8)including combinations of ad hoc (Chapter 4) and cellular networks [6 7] andmobile to mobile(m2m) communications [8] In order to analyze and control these networks evolving towardscomplex networks structures efficient modeling tools are neededComplex network theory (Chapter 14) has emerged in recent years as a powerful tool for

modeling large topologies observed in current networks [9] For instance the World WideWeb behaves like a power-law node degree distributed network wireless sensor networks likelattice networks and relations between social acquaintances like small world networks Theconcept of small world networks was first introduced byWatts and Strogatz [10] where a smallworld network is constructed via rewiring a few links in an existing regular network (such as aring lattice graph) Later on Newman-Watt [11] suggested a small world network constructedby adding a few new links (shortcuts) without rewiring existing links The concept of smallworld can be introduced to wireless networks typically to reduce the path length and thusprovide better throughput and end to end delaySeveral works have addressed the question of how to construct a wireless network topology

in ad hoc and sensor networks (Chapter 5) in such a way that the small world feature is pre-served [12ndash16] Long range shortcuts can be created by adding wired links [17] directionalbeamforming [18] or using multiple frequency channels [19] concepts In Ref [9] it was dem-onstrated that small world networks are more robust to perturbations than other network archi-tectures Therefore any network with this property would have the advantage of resiliency

Advanced Wireless Networks Technology and Business Models Third Edition Savo Glisiccopy 2016 John Wiley amp Sons Ltd Published 2016 by John Wiley amp Sons LtdCompanion website wwwwileycomgoglisic0916

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

where the random omission of some vertices does not increase significantly the average pathlength or decrease the clustering coefficient These features are highly desirable in future wire-less networks where the availability of links and nodes can be uncertain For these reasons inthis book we are interested to redesign heterogeneous wireless networks by including smallworld properties and frequency channels backupsThe considered network model that we envision for 5G and further to 6G includes the

multi-hop concept to model future networks with dense user populations and enablesmobile to mobile (m2m) connections which are already standardized We see multi-hopcellular networks as an extension or generalization of the existing m2m concept Thepotential users acting as relays may belong to different operators and as such may ormay not want to cooperate Consequently the existence of those links will be uncertainSome subareas of the cell will be covered by other technologies such as femto cells smallcells or WLANs enabling the possibility for the cellular system to offload the traffic Theexistence of those links depends on the relaying distance and coverage of the WLAN aswell as the cooperation agreement between the operators In such a complex network cog-nitive links might also be available with limited certainty due to unpredictable activity ofthe primary user (PU) Complex network theory will be used to aggregate all these char-acteristics of the network into a unified model enabling a tractable analysis of the overallsystem performanceDespite of the extensive work in each of the previous fields to the best of our knowledge our

book is the first to provide a unified model of the network that will include simultaneously allthose technologies The dynamic characteristics of the network results into a dynamic networktopology The work developed by [20] represents the first attempt to model the link uncertaintyby complex networks concepts although in this work the uncertainty was a consequence onlyof fading and dynamic channel access More specifically our book emphasizes the followingaspects of the design and analysis of complex heterogeneous wireless networks

1 A unified model for heterogeneous wireless complex networks based on the probabilisticcharacterization of the nodelink uncertainty The model captures the existence of uncertainand time varying links and nodes inherently present in the latest solutions in wirelessnetworks

2 Analytical tools for the unified analysis of the multi-operator collaboration m2m transmis-sion different traffic offloading options and channel availability in cognitive heteroge-neous networks

3 Redesign of heterogeneous networks by using specific techniques to systematically add in acontrolled way network redundancy in order to increase the network robustness to linknode failures

4 Traffic distribution aware rewiring of the heterogeneous network5 A set of new routing protocols for such network6 Comprehensive analysis of the network in terms of average path length clustering robust-

ness power consumption and complexity

In this introduction we start with a general model of the future wireless network referred toas generic network model and later in separate chapters we elaborate in more detail eachcomponent of such network

2 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

11 Network Model

We start by considering a macro cellular network where users transmit uplink by relaying totheir adjacent users (neighbors) on the way to the base station (BS) Multi-hop transmission ismodeled by considering a virtual cell tessellation scheme presented in Figure 111 where themacro cell of radius R is divided into inner hexagonal subcells of radius r lt R This partition isnot physically implemented in the network but rather used to capture the mutual relationsbetween the terminals in the cell that are potentially available for relaying each otherrsquos mes-sages For this purpose it is assumed that if available a potential ready to cooperate transmit-terreceiver is on average situated in the center of each subcellWe assume that within a cell the BS is surrounded by H concentric rings of subcells For the

example in Figure 111 H = 3 The shortest path (in hop count) between the user location andthe BS is given by the hop index h h = 1hellip H Due to the terminal unavailability there maybe routes towards the BS where the length of the path is longer than h The number of subcellsper ring is nh = 6∙h and the number of subcells per cell is N = 3H(H + 1)In the sequel we present a number of characteristics of heterogeneous networks that lead to

the uncertain existence of nodes and links Node percolation will be used to model and quantifythe unavailability of users to relay as a consequence of lack of coverage or terminals belongingto a different operator with no mutual agreement for cooperation When cognitive links areused link percolation is used to model the link unavailability due to the return of the PU tothe channel These options will be elaborated in detail in the subsequent subsections

111 Node Percolation

1111 Multiple Operator Cooperation in Cellular Network

Here we model the scenario where a number of operators coexist in the cellular network It isassumed that a single operator i has a terminal available in a given subcell with probability poi In a multi-operator cooperative network a terminal will be available for relaying in the samesubcell if at least one operator has a terminal at that location This will occur with probabil-ity p = 1minus i 1minuspoi

BS

h = 2 h = 1

dr

h = 3

R

WLAN

Figure 111 Macro cell tessellation

3Introduction

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

This probability is higher for higher number of operators willing to cooperate In general thiswill result into a reduction of the relaying route length If the operators cooperate and let theirusers to flexibly connect to the BS that is more convenient to them the network capacity of bothoperators will be improved Thus a better performance of the network will be obtained in themulti-operator cooperative scenario as will be shown later in this chapter The node unavail-ability for the message forwarding in complex network terminology is referred to as node (orsite) percolation

1112 Multiple Operators in Cooperation with Multiple Technologies

In general multiple technologies will be available in a heterogeneous network Each technologyhas its own characteristics which enables more appropriate AP choice at a specific place andtime based on the usersrsquo requirements Figure 111 shows an example of a cellular networkoverlapping in coverage with a WLAN In the analysis we will be interested to generalize thismodel as follows The relative coverage between the cellular network and other access tech-nologies that is WLAN will be characterized by probability pwlan which is the probability thatin the next hop the connection will have the opportunity to make a handoff to a different tech-nology and so terminate the route The probability pwlan = AA is calculated as the ratio betweenthe coverage areas of other technologies Ah and the coverage area of cellular network Ac Thiscan be easily generalized to introduce other traffic offloading options like smallfemto cells orother multitier elements like micro and pico cells

1113 Modeling m2m Links

In the analysis we will consider the possibility that every next relay on the route will be a finaldestination of an m2m link with probability pm2m This parameter depends on the probabilitythat the session is within the same cell and parameter N representing the number of subcells inthe networkThe simplest model will assume that for a specific session pm2m = (Nm2mN)Nm2m = 1N

where Nm2m is the average number of m2m connections per cell Nm2mN represents the prob-ability that the given adjacent node is a sink for an m2m connection and 1N is the probabilitythat it is a sink for a specific session out of Nm2m such sessions

112 Link PercolationmdashCognitive Links

In the case that cognitive links are used for relaying which means that we are establishing theroutes for the secondary users (SUs belonging to a secondary operator SO) there are tworelated problems that should be considered The first one is the link availability at the momentwhen routingrelaying decision is being made and the second one is the PU return probabilitythat will interrupt the ongoing relaying and force the SU user to try it again with a new optionWe assume that spectrum sensing is perfect [3] Since this problem belongs to the physical

layer technology and has been extensively covered in the literature we will not discuss it withinthis book We also consider that due to the uncertainty of the PUrsquos activities the SO cannotobtain spectrum availability information in advance for the entire message transmission period

4 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

Wemodel this uncertainty by defining a probability of return of the PU to the channel currentlyallocated to the SU denoted as preturnLet us assume that calldata session arrivals follow a Poisson distribution with rate λp and λs

for the PU and SU respectively The average probability pnp that in a given moment np out of cchannels are being used in PO network (the system is in state np) can be obtained as a solution ofbirth death equations for conventional MMc system for data session and MMcc system forvoice applications [21]We assume that the average service time of the SU is 1μs so that the probability of having kp

new PU arriving within that time is [21]

pkp t = 1 μs =λpt

kp

kpeminusλpt =

λp μskp

kpeminusλp μs 1 1 1

The probability that a specific channel among c ndash np channels is allocated to one of the kp newarrivals is kp(c ndash np) So the average corruption probability due to the PU return will be

Pr np =cminusnp

kp = 0

kpcminusnp

pkp t = 1 μs

=cminusnp

kp = 0

kpcminusnp

λp μskp

kpeminusλp μs

1 1 2

The previous expression can be further averaged out over np to give the average PU returnprobability defined as

preturn = npPr np pnp 1 1 3

The models presented so far capture the uncertainty of nodes and links due to differentcharacteristics of wireless networks The network connectivity when all the previous phenom-ena are present in the network is analyzed in the next section by using an absorbing Mar-kov chain

12 Network Connectivity

In modeling network connectivity we will start with the initial model from Figure 111 and allcomponents described in the previous section This initial model is then redesigned later byincorporating the concepts of small world networks and systematic introduction of frequencybackup channels In general we assume that the network is using cognitive links when avail-able If a cognitive link is used and there is a PU return to the channel the ongoing transmissionwill be aborted with probability preturn given by (113) and the user will try another channel Ifthere is no PU return to the channel the user will relay to the receiver of the m2m link if there issuch receiver for a specific session in the neighboring subcells (probability pm2m) This jointevent will happen with probability pm2m (1 minus preturn) Otherwise if there is no such receiver

5Introduction

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

the user will relay to theWLAN if available (probability pwlan) in the neighborhood with prob-ability pwlan(1 minus preturn)(pwlan) In addition to WLAN in general there will be also other optionsfor traffic offloading (smallfemto cells or different tiers of cellular network like pico andmicrocells) The offloading decision will be made with certain probability that depends on a numberof parameters AP availability cost of offloading traffic distribution terminal interface and soon For the purpose of the analysis in this paper all these parameters will be included in pwlanThis is illustrated in Figure 121 If none of these two options is available and there is no returnof the PU the user will transmit towards BS by relaying to the neighboring subcells that willtake place with probability

Pr = 1minuspwlan 1minuspm2m 1minuspreturn 1 2 1

The probabilities of relaying to a specific adjacent subcell are indicated in Figure 121 wherep is the terminal availability probability In each subcell the user checks the adjacent relay thatis in the direction with the shortest distance towards the BSAP The adjacent relay will beavailable with probability p as shown in Figure 121 and if available relaying will take placeas indicated with probability pPr If this user is not available then the protocol checks the avail-ability of the next user in the order indicated in Figure 121 In general the potential relayscloser to the direction of the BS are checked up first More specifically the protocol checksup the right user which will be available with probability p so the probability that this tran-sition will take place is p(1 minus p)Pr In the case of non-availability the protocol will check the leftuser The protocol continues in the same way until it gets to the last adjacent user where relayingwill take place with probability p(1 minus p)5Pr If none of the above options is available then the

3rd

pPr

p(1 ndash p)2 Pr

p(1 ndash p)4 Pr

m2mwlanotheroffloading

options

no route

pwlan(1 ndash pm2m)(1 ndash preturn)

pm2m(1 ndash preturn)

p(1 ndash p) Pr

p(1 ndash p)3 Pr

5th

6th 4th

2nd1st

preturn

p0= Pr(1 ndash pt) p(1 ndash p)5 Pr

Pr= (1 ndash pwlan)(1 ndash pm2m)(1 ndash preturn)

Figure 121 Connectivity alternatives (the direction of the adjacent users is chosen in increasing orderof distance from the BS)

6 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

route will not be established with probability p0 as indicated in Figure 121 As result the rout-ing protocol will be referred to as AP location aware routing Parameter p0 will be used as a keyindicator of the node robustness to link and node failure (unavailability)In general we denote by pn the probability of relaying to adjacent user n obtained as

pn = p 1minusp nminus1Pr n = 1hellip6 1 2 2

where Pr is given by (121) Thus the overall relaying probability to any adjacent subcell isobtained as

pt = npn 1 2 3

In a complex system the simultaneous impact of the number of factors described inSection 11 is included by using the equivalent value of parameter p equal to the product ofthe individual probabilities characterizing the corresponding phenomena For example inthe systemwith two operators with terminal availabilities p1 and p2 respectively the equivalentterminal availability probability is given by

p = peq = 1minus 1minusp1 1minusp2 1 2 4

So the relay will be available if the terminal from at least one operator is available

13 Wireless Network Design with Small World Properties

131 Cell Rewiring

In the previous section the network connectivity is considered from the point of view that theBS is the main target (destination) in the routing protocol This means that most of the traffic isintended for destinations out of the cell In this section we focus our interest on the scenarioswhere most of the traffic remains within the cell and we are primarily interested to improveconnectivity among the nodes within the cell This is typical office scenarios where most ofthe traffic flows between the interoffice computers computers and printers interoffice voiceand video communications and so on Later on we will generalize the network model toinclude multiple cells in the overall complex networkWe start by indexing the subcells along the spiral presented in Figure 131 and unfolding

the spiral into a lattice that will be referred to as s-lattice The lattice obtained this way hassimilar form as those used in the classic literature of the complex networks theory [10 1122 23]In a conventional one-dimensional lattice connections are established between all vertex

pairs separated by k or less lattice spacing The small-world model [10 22 23] is createdby choosing at random a fraction of the edges in the graph and moving one end of eachto a new location also chosen uniformly at random In a slight variation on the model in[10 11] shortcuts are added randomly between vertices but no edges are removed from theunderlying one-dimensional lattice

7Introduction

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

One can see that in s-lattice obtained by unfolding the spiral from Figure 131 each vertexis connected to six neighbors Different from the conventional lattice the adjacent neighbors onthe spiral are not adjacent neighbors in the lattice any more This fact for itself brings the elem-ents of rewiring or adding additional short cuts More precisely one can see in Figure 131 (lefthand side in shade) that each node in the h-th round of the spiral is connected to two adjacentnodes (k = 1) with the same h two adjacent nodes on the (h minus 1)-th round of the spiral and twoadjacent nodes on the (h + 1)-th round of the spiralIf the coverage of transmission is extended to include two layers of subcells (lattice range

k = 2) around each node (see the right hand side of Figure 131 in shade) then each node inthe h-th round of the spiral is connected to four adjacent nodes with the same h four adjacentnodes on the (h plusmn 1)-th round of the spiral and three adjacent nodes on the (h plusmn 2)-th round of thespiral One should notice that for the nodes located at the corners of the spiral (θ = 30 + 60n n =1hellip6 with respect to the BS) the size of the clusters at the rounds h +Δh and h ndashΔh are notequal This is illustrated in Figure 132 for nodes 2 and 3 of the spiral in Figure 131Formally parameter k for s-latticemeans that each node will be connected to the 2k + 1 clus-

ters located on adjacent rounds of the spiral within distance Δh le k with each individual clustersize le kLet us denote by u(h θ) the user (network vertex) located in hop h and angle θ In vector

representation its location is given as u hθ = h dr ejθ where dr is the relaying distanceThe locations of its adjacent relaying users connected for certain lattice range k are given inthe Appendix A1 The s-lattice with shortcuts will be referred to as s(sc)-lattice

BS

1

2

3

4

5

6

7

89

10

1135

Figure 131 s-Lattice parameters

8 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

132 Traffic Distribution Aware Rewiring

It is intuitively clear that from the routing and delay point of view we will need a shortcutbetween nodes with high traffic density On the other hand a direct link between nodes faraway from each other would require high power to maintain it In order to accommodate thesecontradictory requirements we suggest a traffic distribution aware rewiring where the shortcutsare established following one of the options provided below with probability

pij λij 1 3 1

By considering the power consumption (131) can be modified as

pij λij Pij 1 3 2

or equivalently

pijλijPij lePthreshold

0Pij gtPthreshold1 3 3

These probabilities may be also obtained as a solution of the more sophisticated optimizationproblem with more complex utility functionIn practice the shortcuts can be implemented by using separate m2m channels from the

macrocell or equivalently by considering channel reuse factor 1 and scheduling the transmis-sions in different slotsIn the case of rewiring referred to as s(r)-lattice the rewired link will be removed and recon-

nected randomly to another node For both the s(sc)- and s(r)-lattices a new set of protocolswill be developed later

123

5

4

NN-1

6

7

8

9

1110

123

5

4

NN-1

6

7

8

9

1110

12

(a) (b)

Figure 132 s-Lattice connection model for (a) user 2 and (b) user 3

9Introduction

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

133 Multicell Rewiring

Multiple cells can be interconnected by using two way spiral 2ws-lattice with 2N nodes asshown in Figure 133 The rewiring (or adding shortcuts) is performed between the two ran-domly chosen nodes from the whole network Physically this can be implemented by using thenetwork backholes and direct link (macrocell orWLAN) from the nodes to the nearest backholeaccess point for rewiring

BS

1

2

3

4

5

6

7

89

10

11

BS

2N

2N-12N-2

Figure 133 2ws-Lattice

10 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

14 Frequency Channels Backup

In this section in addition of the small world properties we consider the possibility that a num-ber of additional channels (either cognitive or purchased licensed channels) will be available forrelaying There are a number of ways how additionally purchased licensed channels can bemade available to increase the overall network robustness to the link and node failure ThePO can sell the channel with respect to

Area (A sell)per macro cellper constalletion unit (subcell)

Number of frequency channels (F sell)one (1) orkf channels

Time the contract is valid (t sell)temporal (per session) orfixed time sell

In the sequel we will use AFt notation for an A sell F sell t sell contract As an example amkf s contract refers to the sell on the macro cell area kf channels for the duration of a givensession Depending on the type of the sell different effects will be achieved with respect to thenetwork robustness enhancement

141 mkfs Contract

We characterize the network state with (npns) where np is the number of temporally active usersin the primary network and ns is the same parameter for the secondary network For a givenoverall number of available channels c PO will keep bp channels as its own backup and is readyto temporally sell to SO c ndash (np + bp) channels The SO will buy bs channels for its own back upand the rest of the free channels will be used as cognitive channels Parameter bs is limited tobs lt kf and can be represented as

bs =kf cminus np + bp ge kf

cminus np + bp cminus np + bp le kf1 4 1

142 Random Redundancy Assignment (R2A)

In this case the backup channel is randomly assigned to ns users resulting in backup probabilityin secondary network defined as pbs = bs ns

11Introduction

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks

143 On Demand Redundancy Assignment

In this case the redundant channel is assigned to the terminal after s successive returns of PU tothe channel This can be modeled as

p1 = psreturn 1 4 2

pi =ns

ipi1 1minusp1

ns minus i 1 4 3

pbs =kf minus1

i= 0

pi 1 4 4

where (142) defines the probability that s successive returns have occurred after which thesubcell demands for a backup channel Parameter pi is the probability that out of ns activeSU i users are using the backup channel Finally (144) defines the probability that at leastone out of kf leased channels is free to be allocated to the new demand The optimum value ofparameter s is obtained as

s= argmaxs

pbs s

= argmaxs

1s

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5a

Equation 145a searches for the value of s that maximizes the probability that at least one outof kf leased channels is free to be allocated to the new demand For higher s SUs will need towait longer and hope that there will be no additional returns of the PU so that they can finallytransmit without asking for the backup channel It is intuitively clear that higher s will reducethe probability of having i SUs needing backup channels which is defined by (143) and thusincrease the probability once the backup channel is requested that there will be a backup chan-nel to meet such a request as given by (144) On the other hand we cannot allow s to be toohigh since this will increase the overall delay of message delivery to the access point Thereforethe utility function in (145a) is obtained by dividing Pbs by s This utility function can befurther modified to obtain s as

s= argmaxs

pbs slr

= argmaxs

1slr

kf minus1

i= 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5b

s = argmaxs

pbs sl

= maxs

1sl

kf minus1

i = 0

ns

ipisreturn 1minuspsreturn

ns minus i1 4 5c

12 Advanced Wireless Networks