1 3SPRI NGERBRI EFSI NCOMPUTERSCI ENCENingZhangJon W.MarkSecurity-aware Cooperation in Cognitive Radio NetworksSpringerBriefs in Computer ScienceSeries EditorsStan ZdonikPeng NingShashi ShekharJonathan KatzXindong WuLakhmi C. JainDavid PaduaXuemin ShenBorko FurhtV.S. SubrahmanianMartial HebertKatsushi IkeuchiBruno SicilianoFor further volumes:http://www.springer.com/series/10028Ning Zhang Jon W. MarkSecurity-aware Cooperationin Cognitive Radio Networks1 3Ning ZhangDepartment of Electricaland Computer EngineeringUniversity of WaterlooWaterloo, ON, CanadaJon W. MarkDepartment of Electricaland Computer EngineeringUniversity of WaterlooWaterloo, ON, CanadaISSN 2191-5768 ISSN 2191-5776 (electronic)ISBN 978-1-4939-0412-9 ISBN 978-1-4939-0413-6 (eBook)DOI 10.1007/978-1-4939-0413-6Springer New York Heidelberg Dordrecht LondonLibrary of Congress Control Number: 2013958216 The Author(s) 2014This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part ofthe material is concerned, specically the rights of translation, reprinting, reuse of illustrations, recitation,broadcasting, reproduction on microlms or in any other physical way, and transmission or informationstorage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodologynow known or hereafter developed. Exempted from this legal reservation are brief excerpts in connectionwithreviewsorscholarlyanalysisormaterial suppliedspecicallyforthepurposeofbeingenteredandexecutedonacomputersystem, forexclusiveusebythepurchaserofthework. Duplicationofthispublicationorpartsthereof ispermittedonlyundertheprovisionsoftheCopyright LawofthePublishers location, in its current version, and permission for use must always be obtained from Springer.Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violationsare liable to prosecution under the respective Copyright Law.The use of general descriptive names, registered names, trademarks, service marks, etc. in this publicationdoes not imply, even in the absence of a specic statement, that such names are exempt from the relevantprotective laws and regulations and therefore free for general use.Whiletheadviceandinformationinthis bookarebelievedtobetrueandaccurateat thedateofpublication,neither the authors nor the editors nor the publisher can accept any legal responsibility forany errors or omissions that may be made. The publisher makes no warranty, express or implied,withrespect to the material contained herein.Printed on acid-free paperSpringer is part of Springer Science+Business Media (www.springer.com)PrefaceCognitive radio networks (CRNs) are envisaged to solve the problem of spectrumscarcity, caused by thecurrent spectrum allocation policy inwhich only licensedusers have channel access rights. In CRNs, unlicensed users are allowed toopportunisticallyusetheidlespectrumbandsownedbylicenseduserstomeetthe ever-increasing demand on spectrum and increase spectrum efciency. In ordertoaccessthespectrumbandswithout creatinginterference tothelicensedusers,unlicensedusersneedtoconductspectrumsensing. However, spectrumsensingmight be inaccurate due to multipath fading, shadowing, and primary receiver uncer-tainty. To address this problem, two types of cooperation have been proposed in theliterature: cooperative spectrum sensing and cooperative cognitive radio networking(CCRN). Fortheformer,thecooperationisperformedamongunlicensedusers;whileforthelatter, thecooperationiscarriedoutbetweenunlicensedusersandlicensed users. As an emerging paradigm, CCRNcan achieve mutual benets forboth participants, which will be the focus of this book. Whereas cooperation canalso incur security issues, e.g., malicious users might participate in the cooperationto corrupt or disrupt the communication of legitimate users. Those security issuesare of great importance and need to be addressed before the widespread deploymentof cooperation in CRNs.In this book, we study cooperative networking in CRNs, where unlicensed usersandlicensedusers cooperatewitheachother toobtainmutual benets, takingsecurityaspectsintoconsideration.InChap. 1, werst giveabriefintroductionto CRNs, including fundamentals of cognitive radio, spectrum sensing, and cooper-ation in CRNs. In Chap. 2, a literature survey on cooperative networking in CRNs isprovided, followed by a discussion on the security aspects, which also motivate thesubsequent works in Chaps. 3 and 4. Specically, a trust-aware cooperation schemefor CRNstoimprove throughput or energy efciency oflicensed users and offertransmission opportunities tounlicensed users, considering the trustworthiness ofunlicensedusers, ispresentedinChap. 3; andacooperationschemetoenhancesecure communications of licensed users is presented in Chap. 4. Numerical resultsare provided also to evaluate the proposed schemes. Finally, the concluding remarksare given in Chap. 5.vvi PrefaceThe authors wouldlike tothankProf. Xuemin(Sherman) Shen, NingLu,andNanChengoftheBroadbandCommunicationsResearchGroup(BBCR)attheUniversityof Waterloo, andProf. RongxingLuof NanyangTechnologicalUniversity, for their contributions in the presented research works. Special thanksare also due to the staff at Springer ScienceBusiness Media: Courtney Clark andJennifer Malat, for their help throughout the publication preparation process.Thanks also to the rst authors parents, Jianguang Zhang and Jianhua Duan, andhis wife, Xue Qin, for their love and support.Waterloo, ON, Canada Ning ZhangWaterloo, ON, Canada Jon W. MarkContents1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Cognitive Radio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.1.2 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.1.3 Network Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.1.4 Operational Modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.2 Spectrum Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.2.1 Energy Detection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.2.2 Cyclostationary Feature Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.2.3 Matched Filter Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.2.4 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.3 Cooperation in CRNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.3.1 Cooperative Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.3.2 Cooperative Networking in CRNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Cooperative Cognitive Radio Networking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1 Literature Survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1.1 Network Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1.2 Cooperation Setting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.1.3 Cooperation on Different Layers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.1.4 Cooperation Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.1.5 Cooperation Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.2 Security Aspects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.1 Trust-Aware Cooperation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.2 Cooperation for Secrecy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22viiviii Contents3 Trust-Aware Cooperative Networking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.2 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.2.1 MAC Layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243.2.2 Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253.3 Cooperation for Throughput Improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.3.1 Trust Computational Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.3.2 Stackelberg Game between PU and SU. . . . . . . . . . . . . . . . . . . . . . . . 273.3.3 Game Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.3.4 Existence of the Stackelberg Equilibrium . . . . . . . . . . . . . . . . . . . . . . 333.3.5 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.4 Cooperation for Energy Saving. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.4.1 Stackelberg Game between PU and SU. . . . . . . . . . . . . . . . . . . . . . . . 373.4.2 Game Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.4.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424 Cooperative Networking for Secure Communications. . . . . . . . . . . . . . . . . . . . 434.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434.2 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454.3 R-J Cooperation Scheme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464.3.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464.3.2 Cooperation Parameters Determination. . . . . . . . . . . . . . . . . . . . . . . . 494.4 C-B Cooperation Scheme with E-CSI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514.4.1 C-B Scheme for Single Eavesdropper (CBSE) . . . . . . . . . . . . . . . . . 524.4.2 C-B Scheme for Multiple Eavesdroppers (CBME) . . . . . . . . . . . . 554.5 C-B Cooperation Scheme Without E-CSI (CBNE). . . . . . . . . . . . . . . . . . . . 584.5.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584.5.2 Cooperation Parameters Determination. . . . . . . . . . . . . . . . . . . . . . . . 594.6 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71AcronymsAF Amplify-and-forwardBS Base stationC-B Cluster-beamforming cooperation schemeCBSE Cluster-beamforming scheme for single eavesdropperCBME Cluster-beamforming scheme for multiple eavesdroppersCCRN Cooperative cognitive radio networkingCR Cognitive radioCRNs Cognitive radio networksCSI Channel state informationDF Decode-and-forwardDSA Dynamic spectrum accessDSTC Distributed space-time codingE-CSI Eavesdroppers channel state informationFDMA Frequency-division multiple accessMRC Maximal ratio combiningNE Nash equilibriumPDF Probability density functionPHY Physical layerPU Primary userQoS Quality of serviceR-J Relay-jammer cooperation schemeSU Secondary userSNR Signal-to-noise ratioTDMA Time division multiple accessixChapter 1IntroductionAbstract Cognitive radio is a promising solution to the spectrum scarcity versusunderutilization dilemma, which enables unlicensed users to opportunistically usethe unused spectrum owned by licensed users to increase the spectrum utilization.To avoid interfering with the operation of licensed users, unlicensed users performspectrum sensing before transmission to detect the available channels. However, theoutcome ofspectrum sensingmaybeinaccurate duetofadingorshadowing. Toovercome this problem, cooperation has been leveraged in cognitive radio networks(CRNs). Inthis chapter, werst giveabrief introductiontoCRNs, includingfundamentals of cognitive radio, spectrum sensing, and cooperation in CRNs.1.1 Cognitive RadioIn recent decades, it has been witnessed a rapid growth in wireless communications,whichhavebeen almostapplied toevery aspects ofpersonal. Emerging wirelessdevicesandapplicationsfurtheracceleratethedevelopment ofwirelesssystems.Such an exponential growth of wireless communication also imposes huge demandsonradiospectrum. Asanatural resource, radiospectrumisscarceandlimited.Nowadays, the spectrum is managed by government agencies such as the FederalCommunications Commission (FCC), and assigned to licensed users on a long termbasis to avoid interference among wireless systems. Although this static allocationapproachworkedwell inthepast, it cannot servetheever increasingdemandfor wireless communicationwell becauseof theproblemof spectrumscarcity.Recent studies reveal that theallocatedspectrumareunderutilized. As showninFig. 1.1, somepartsofspectrumremainlargelyunderutilized,somepartsaresparingly utilized, while the remaining parts of the spectrum are heavily occupied[1].It isrecognized thatthis kind ofstaticallocation policy hasresulted inpoorspectrumutilization, and created a severe shortage of spectrumfor unlicensed users.Furthermore, spectrumunderutilizationbylicensedusers exacerbates spectrumscarcity. The main reason of spectrumunderutilizationis that licensed usersN. Zhang and J.W. Mark, Security-aware Cooperation in Cognitive Radio Networks,SpringerBriefs in Computer Science, DOI 10.1007/978-1-4939-0413-6__1, The Author(s) 201412 1 IntroductionMaximum AmplitudesFrequency (MHz)Amplitude (dBm)Heavy UseSparse Use Medium UseHeavy UseFig. 1.1 Spectrum utilization [1]typically do not fully utilize their allocated bandwidths for most of the time, whileunlicensedusers arebeingstarvedfor spectrumavailability. Todeal withthisdilemma, cognitive radio is a paradigm created in an attempt to enhance spectrumutilization, by allowing unlicensed users to coexist with licensedusers and makeuse of the spectrum holes [13]. The spectrum holes are dened as the spectrumbands owned by licensed users, which are unused at a particular time and specicgeographic location.Cognitive radio (CR) is dened as a radio that can sense the surrounding wirelessenvironments where it operates and adjust the transmission parameters accordingly.Tobemoreprecise, FCCgives thedenitionas follows: Cognitiveradio: Aradioorsystemthatsensesitsoperational electromagneticenvironment andcandynamicallyandautonomouslyadjust itsradiooperatingparameterstomodifysystemoperation, suchasmaximizethroughput, mitigateinterference, facilitateinteroperability, access secondary markets. [4].Two main characteristics that distinguish CR from the traditional wireless radioare cognitive capability and recongurability. The former represents the awarenessof CRwithrespect tothetransmittedwaveform, RFspectrum, communicationnetwork, geography, locally available services, user need, security policy and so on,while the latter corresponds to capability of adaption to the obtained informationabout the wireless environments [3].Tobetterunderstandcognitiveradio, software-denedradio(SDR) isbrieyreviewed. Introduced in 1991, SDR is dened as a radio platform, whereby compo-nents implemented in hardware (modulation/demodulation, compression, ltering,errorcorrectioncoding,etc.)canbeimplemented bymeansofsoftwareinstead.Thanks to the rapid development of microelectronic, wireless devices become moreandmorepowerful andcapable,whichinreturn facilitatetheevolution ofSDR.1.1 Cognitive Radio 3WithSDR, thefunctionalitiesor theoperational parametersof thedevicescanbeprogrammable, whichenablesrecongurationof theradiobyjust changingthecodes. Withthefeatureofrecongurability,manynewapplicationsemerge.Forinstance, SDRcanbeusedtobuildradiostosupportmultipleinterfacesbyreconguring them in software. In addition, the base transceiver station (BTS) orother devices, implemented using SDR, can be upgraded simply without too muchexpense, e.g., from Global System for Mobile Communication (GSM) to Generalpacket radio service (GPRS).AlthoughSDRcanprovidetherecongurablecapability, it canonlybeper-formedondemand. Inotherwords,SDRcannotcarryoutreconguration itself.Different fromSDR, cognitive radio can achieve self-conguration through learningfromtheenvironment whereit operatesduetothefundamental characteristicsofcognitivecapabilityandrecongurability. Withcognitivecapability,cognitiveradio can acquire theinformation from theenvironment and adapt without beingprogrammed a priori. With the recongurability, cognitive radio can be dynamicallyprogrammed accordingly, which is implemented on the platform of SDR. In otherwords, cognitive radio can automatically sense or detect the radio environment, andchange its transmission or reception parameters accordingly such that the resourcecan be utilized in an efcient way.1.1.1 FunctionsWith CR technology, unlicensed users can coexist with licensed users to exploit thespectrum holes. However, there is a stringent requirement on the unlicensed users.That is, the operation of unlicensed users must not interfere with the transmissionsof licensedusers. Moreover, consideringthe diverse qualityof service (QoS)requirements of unlicensed users,CRfaces many challenges. Toovercome thosechallenges, the following functions of CR are identied: spectrumsensing, spectrumdecision, spectrum sharing, and spectrum mobility [5].Spectrum sensingisavery important function, whichshould beperformed toacquireinformationfromthesurroundingenvironment, suchaspresenceofthelicensed users and channel availabilities, before transmission [6, 7]. It is necessaryfor CR to adapt its operational parameters according to the status of the environment.The objectives of spectrum sensing can be classied as follows: (1) the operationofunlicensed users should avoidharmful interference tolicensedusersbyeitherswitchingtoanavailableband orlimitingitsinterference tolicensedusers atanacceptablelevel,and(2)unlicensedusersshouldefcientlyandreliablyidentifythe spectrum holes to meet their QoS requirements. Therefore, spectrum sensing iscrucial for both licensed and unlicensed users.After available channels are detected, spectrum decision is performed to selectsuitable channels accordingto the QoSrequirement of unlicensedusers. Thedecision is made based on the results of spectrum sensing and the internal policy of4 1 Introductionthe users (e.g., to maximize throughput, reliability, or have the longest transmissiontime, andsoon). Subsequently, thebest channel toaccess is selectedamongavailable channels.When multiple unlicensed users exist, spectrum sharing is necessary, especiallyfor distributed CRNs. Spectrum sharing refers to the process of sharing the commonavailable channels among multiple unlicensed users. The objective is to utilize theavailable channels in an efcient and fair way by coordinating the users [810].Since an unlicensed user has to vacate the current channel once the presence oflicensed user is detected, the unlicensed user has to nd another available channelto access in order to maintain the ongoing transmission. This process is referred toas spectrum mobility. The goal is to meet the QoS requirement of unlicensed usersby means of choosing the channel to move or sense.1.1.2 ApplicationsSinceCRiscapableofautonomouslyadaptingitsoperationalparameters(e.g.,transceiver parameters) to work in a more efcient way, based on the informationacquired from the environment by active monitoring, a large number of promisingapplications can be facilitated, ranging from military to commercial market. Twokeyapplications,dynamicspectrumaccess[1115]andinteroperability [16, 17],are identied in the academia and industry. Dynamic spectrum access refers to thescenario where unlicensed users sense the available channels which are not occupiedby licensed users, and then access those channels for transmission. Interoperabilitymeans that radios can connect different systems operating on different protocols orstandards so that they can communicate with each other.1.1.2.1 Dynamic Spectrum AccessDynamicspectrumaccess(DSA), asshowninFig. 1.2, isthemainapplication,which has received great attention from both academia and industry. The objectiveof DSAistoefcientlyutilizethespectrumtosolvetheproblemof spectrumsecrecy, which is the result of the ever increasing mobile devices and the currentstatic spectrum allocation policy. DSA allows unlicensed users to opportunisticallyutilizethelicensedspectrumbands whentheyareunoccupied. Inordertoavoidharmful interferencetothe legacysystem, unlicensedusers havetocarryoutspectrumsensingtodetect thespectrumholes. Oncetheavailablechannelsaredetected, unlicensed users can access for their transmission. During transmission,spectrumsensinghas tobe continuouslyperformedtosense the activities oflicensed users. When the presence of a licensed user is detected, the unlicensed uservacates thecurrent channel and chooses other channels tosensefor transmissionopportunities.1.1 Cognitive Radio 5a bFig. 1.2 Dynamic spectrumaccess. (a) Wireless systems without CR, (b) wireless systemswith CRa bCognitiveRadioFig. 1.3 Interoperability enabled by cognitive radio. (a) Wireless systems without CR,(b) wireless systems with CR1.1.2.2 InteroperabilityAs the second important application, interoperability has ahuge potential impactonthecurrent communicationarchitecture, whichis expectedtofurther affectthepersonallifeofhuman. Nowadays, wearesurroundedbydifferenttypesofcommunication systems, e.g., mobile networks, sensor networks, wireless local areanetwork, TVbroadcastnetwork, andsoon. Thosesystemsareindependentandautonomous systems, with different standards, spectrum bands, services, etc. Withthe technology of cognitive radio, the device can recongure itself to communicatewith incompatible radios. Specically, CR rst scans the surrounding environmentto detect what waveforms or networks are present. Then, it can either recongureitself to communicate with the selected network, or it can act as a gateway to connectdifferent systemsforcommunications. Bydoingso, diversewirelesssystems,asshown in Fig. 1.3, can be connected and communicate with each other.1.1.3 Network ArchitectureWith the assistance of cognitive radio technology, unlicensed users can coexist withlicensed users and utilize the temporarily unused spectrum bands owned by licensedusers. Therefore,CRnetworkarchitectureiscomprisedoftwocomponents:theprimary network and the secondary network, as shown in Fig. 1.4. Both networks6 1 IntroductionUnlicensedBandLicensedBand ILicensedBand IIPrimaryNetworkDistributedSecondaryNetworkInfrastructuredSecondaryNetworkSPECTRUM BANDSecondaryUserPrimary UserPrimary UserPrimary BaseStationPrimaryNetworkAccessCRNetworkAccessCR AdHocAccessSecondaryUserSecondaryBaseStationOtherCRNsSpectrumBrokerFig. 1.4 Cognitive radio network architecture [5]can be deployed in either a centralized or ad hoc mode, where communications arecoordinated by central nodes such as base stations or communications are carriedout in a peer-to-peer fashion, respectively. In cognitive radio networking, unlicensedusers are referred to as secondary users (SUs), while licensed users are coined to asprimary users (PUs).1.1.3.1 Primary NetworkThe primary network corresponds to an existing network which holds a license foroperationincertainspectrumbands.Thisnetworkhastheexclusiveprivilegetoaccess the assigned spectrum bands. If the primary network has an infrastructure,PUscanbecoordinatedtoaccessthenetworkthroughtheprimarybasestation.In addition, the primary network might bedeployed inad hoc mode, where PUscommunicate with each other without any infrastructure. The PUstransmissionsoccurring in the primary network should be protected frombeing interferedbysecondarynetworks. Generallyspeaking, PUsandprimarybasestationsaretypically not equipped with CR functions. Therefore, it is the responsibility of SUsto sense the channel before transmission and vacate the occupied channel when PUsre-appear.1.1 Cognitive Radio 71.1.3.2 Secondary NetworkThesecondarynetwork, composedof aset of SUs, does not havethelicensetooperateinanylicensedspectrumbands. Thesecondarynetworkcanalsobeclassied into twotypes: infrastructure-based and ad hoc [18]. An infrastructure-based secondary network has a central controller, e.g., a secondary base station oran access point. Opportunistic spectrum access by SUs is usually coordinated by thecentral controller. Whereas in an ad hoc secondary network, SUs can communicatewith each other via multi-hop wireless links on either the licensed or the unlicensedspectrumbands. BothSUs andsecondarybasestations areequippedwithCRtechnology. When different secondary networks share one common spectrum band,a spectrum broker is needed to coordinate them.1.1.4 Operational ModesThe operational modes include interleave, underlay, and overlay [19], asdetailedbelow.1.1.4.1 InterleaveFor interleave mode, SUs has to carry out spectrum sensing todetect the unusedchannels before transmission. Only when there is no active PU in the channel, cananSUaccessfortransmission.Duringtransmission,spectrumsensinghastobeperformed tocheckwhetherornotthePUhasreturned. OncethepresenceofaPUisdetected, theSUshould vacatethecurrent operating channel andsensetond another idle channel for transmission. This mode has the disadvantage of beingsensitive to detection errors and the PUs activities.1.1.4.2 UnderlayThe underlay mode allows the SU to transmit data simultaneously with the PU underthe condition that the interference caused by the SU at the primary receiver shouldbebelowapredenedthreshold.Itisdifferentfromtheinterleavemode,wheretheinterference iscompletely avoided. However, this mode relies on anaccurateestimateoftheinterferingchannel, whichinrealityishardtoobtain.Moreover,theSUhaspoorperformancebecauseof thefollowingreasons: (1)duetotheinterferencelimit, theSUhasapowerconstraintwhichaffectsitsperformance;(2) the SUs transmission also suffers from interference by the PUs transmission.8 1 Introduction1.1.4.3 OverlayThe overlay mode leverages cooperative networking to exploit transmissionopportunities, whereby the SUcooperates with the PUto enhance the PUstransmissionperformanceintermsof throughput, reliability, andsoon, andinreturnthePUgrantsaintervaloftimetotheSUforitsowntransmission.Thismode is actually based on mutual benets, i.e., both participants get benets fromcooperation. In the literature, this mode is also called cooperative cognitive radionetworking (CCRN), which is also the focus of this book.1.2 Spectrum SensingTraditional, spectrum sensing is an essential component of cognitive radio to exploitthe spectrumbands opportunistically. Theobjectiveof spectrumsensingis tocheckthechannel availabilityinorder not toadverselyaffect theperformanceof PUs [20, 21]. Sincespectrumholescanbeinspecictime, or afrequencyband, or at aspatial location, spectrumsensingcanbeperformedinthetime,frequency, andspacedomains.Althoughthemainjobofspectrumsensingistoobtain channel availability information, it can also be used to determine the types ofsignals occupying the spectrum, which may include modulation, carrier frequency,waveform, bandwidth, etc.SUs perform spectrum sensing before commencing transmission to avoid inter-ferencetoPUs. Particularly, theSUscansacertainspectrumrangeanddetectsthe spectrum hole, and then accesses the channel for its transmission. During thetransmission, if the SU detects the presence of PUs, it must refrain from utilizingthat band and searches for a newband. In the literature, popular detection techniquesinclude energy detection, cyclostationary detection, and matched ltering [2124].1.2.1 Energy DetectionEnergydetectionis basedonthefact that theenergyof thesignal is usuallylarger thanthatofnoise.Todetermine theexistenceofPUs,theenergy detectorcompares its output (e.g, the average or the total energy of the observed samples)with a predened threshold, which is derived based on the statistic of noise. If theoutput isabovethethreshold, thentheenergydetectormakesthedecisionthatthe PUis present; otherwise, it makes the decision that the PUis absent. Energydetection is the most common type of spectrum sensing technology because of thefollowing reasons: (1) it is simple to implement; (2) it does not require any a prioriinformationregardingthePUs signal; (3)thedetectiontimeisrelativelyshort[2528]. To evaluate the performance of the detector, the probability of detectionand the probability of false alarm are used, which are dened by the probabilitiesthat the output of the detector is above the threshold given that the PU is actually1.2 Spectrum Sensing 9present, andtheoutput ofthedetectorisabovethethresholdwhenthereisnoPU present, respectively. As a good detector, it should have a high probability ofdetection and a low probability of false alarm.1.2.2 Cyclostationary Feature DetectionTypically, there are certain inherent features associated with the signal transmittedby PUs, which can be exploited to detect the presence of PUs. Considering that formost communication systems the signals are cyclostationary due to the periodicityin the signals or the statistics, while the noise is usually assumed as a wide-sensestationary process without correlation; the cyclostationary features can be leveragedto distinguish the PUs signal and noise [29,30]. Through the cyclostationary featuredetection,featuresofPUssignalcanbeextractedtodeterminetheexistenceofPUs. Compared with energy detection, cyclostationary feature detection can providebetter performance for the scenario of low SNR, with the price of high complexity.Moreover, a priori knowledge regarding the characteristics of PUs signal is needed.1.2.3 Matched Filter DetectionIn most wireless systems, pilot bits are periodically transmitted for channel estima-tion, synchronization, and so on. The pilot bits are public information, which can beused to detect the presence of PUs. When the knowledge about the transmitted signalis available in the rst place, matched lter detection will be the optimal detectionapproach, because it can correlate the received signal with the known primary signalforthedetection.Thematchlterhastheadvantage ofshortdetection timeandit workswell inthelowSNRregime. But it requiresperfect knowledgeofthecharacteristics of PUs signals, e.g., modulation type, bandwidth, center frequency,etc. Any imperfect information about the PUs signal will lead to severe degradationin the detection performance.1.2.4 LimitationsSpectrumsensingiscriticalfortheoperationofCR. However,theperformanceofsensingislimitedbyseveral factors, includingmultipathfading, shadowing,primary receiver uncertainty problem [31]. When the SU is experiencing multipathfading or shadowing, thereception ofPUssignal willbesignicantly degraded,which adversely affects the detection accuracy. In addition, for SUs which are outof thetransmissionrange of theprimary transmitter, theycannot detect thePUstransmission. Therefore, when those SUs start to transmit, harmful interference willbe created at the primary receiver, if the primary receiver is unfortunately located10 1 IntroductionPU1PU2SU1SU2SU4SU3Fig. 1.5 Limitations of spectrum sensingwithinthetransmissionrangeof theSUs, whichgivesrisetoprimaryreceiveruncertainty problem. AsillustratedinFig. 1.5,when1U1istransmitting datato1U2, SU1 can receive signal of 1U1 and know the presence of PUs. However, SU2cannot detect 1U1 because the building blocks the signal from1U1. For SU3, sinceit is outside of the transmission range of 1U1, it cannot detect the PUs transmissionandthereforeitstartsitsowntransmission, whichwillcauseinterferencetotheprimary receiver, i.e., 1U2. Moreover, spectrum sensing consumes energy to detectthe spectrum holes and has to be continuously carried out during the transmissionto detect PUss activities.1.3 Cooperation in CRNsTo overcome the limitations of spectrum sensing, cooperation has been introducedinCRNs, whichhas twoforms: cooperativespectrumsensing(CSS) [3234]and cooperative cognitive radio networking (CCRN) [3537]. For the former, thecooperationiscarriedoutamongSUs, wheremultipleSUscooperatewitheachother to enhance the detection performance. For the latter, the cooperation is carriedout between SUs and PUs, where SUs can gain transmission opportunities throughcooperation with PUs. As a promising paradigm, CCRN can relieve SUs from theburden of spectrum sensing. Moreover, it can solve the issue of dynamics of SUstransmission caused by the PUs activities. Therefore, we mainly focus on CCRNin this book.1.3.1 Cooperative SensingCooperative spectrum sensing that relies on spatial diversity and multiuser diver-sitycanimprovethedetectionperformanceintermsofincreasingthedetectionprobability and reducing the false-alarm probability [38, 39], as shown in Fig. 1.6.Instead of using individual decision, multiple SUs share the sensing results to make1.3 Cooperation in CRNs 11SU3SU2SU1PU2PU1Fig. 1.6 Cooperative sensingacombineddecisionthroughcooperation. Particularly, eachSUperformslocalsensing and reports the detection results to a fusion center to make a nal decisionin a centralized fashion, or exchange the local detection results among themselvesina distributedfashion. Throughcooperation, SUs share their sensingresultsandmakeacombinedcooperativedecisionderivedfromthespatiallycollectedobservations, which can overcome the deciency of individual observations at eachSU.Ithasbeenshownthatcooperative spectrum sensingcaneffectively combatmultipathfadingandshadowing, mitigatethereceiveruncertaintyproblem, andhence signicantly improve the detection performance [22, 32, 33, 40].Typically, for thecentralizedcooperativespectrumsensing, it is carriedoutfollowingathree-stepprocess. First, individualSUsperformlocal sensingsep-arately. Then, all thecooperatingSUsforwardthesensingresultstothefusioncenter, whichmight bethebasestation,acommon receiver andsoon.Last,thefusion center combines all the received sensing results and makes a nal decisionon whether the PUis present or absent on the observed band. For the distributedcooperative spectrum sensing, where there is no fusion center, SUs exchange thedetectionresultsamongthemselvesandthenconvergetoanal decisionafterseveral iterations.Many works on cooperative spectrum sensing have been reported in the literature[32, 33, 4143]. The authors in [32] propose a cooperative spectrum sensing schemetoimprove thespectrumsensinginthepresenceofshadowing orfadingeffects.In [42], the authors propose a relay-based cooperation mechanism, which is a two-user cooperative spectrum sensing scheme. This cooperation scheme shows that thedetection time can be reduced. The authors in [41] propose a selective-relay basedcooperative sensing scheme with no dedicated reporting channel. In [43], they alsostudy the sensing and transmission trade-off and show that the performance in termsofthespectrumholeutilizationcanbesignicantlyimprovedusingcooperativerelaying. Anoptimal sensingschemeforthemultiusercooperationisproposedin [33].If all theparticipatingSUs behavewell, thedetectionperformancecanbeimprovedthroughcooperativesensing. However, inanuntrusted, evenhostileenvironment, malicious users might launch different attacks to disturb the detection,12 1 IntroductionhencejeopardizetheoperationsinCRNs. For instance, malicious users mighttransmit signals with characteristics similar to those of PUs or just send jammingsignals to the target channel to interfere with the sensing process and signicantlyreduce the throughput of legitimate SUs. The former is usually referred to as primaryuser emulation (PUE) attack, while the latter is called jamming attack. Moreover, themalicious user might send false sensing report to the fusion center, so as to misleadthe spectrum sensing results and severely degrade the performance of cooperativesensing [40], which is called false sensing report attack. In [44], the authors showthat the PUE attack can signicantly reduce the available resources to the SUs andpropose a transmitter verication scheme to combat the PUE attack. To deal with thefalse sensing report attack, the authors in [45] propose a robust distributed spectrumsensingtoachievearelativelyaccuratesensingdecisioninspiteofthat attack.While the authors in [46] propose an outlier detection to pre-lter the extreme valuesfrom sensing data. To mitigate the control channel jamming attack, the authors in[47] propose a randomized distributed scheme based on frequency hopping, whichallows SUs to establish a new control channel.1.3.2 Cooperative Networking in CRNs1.3.2.1 Cooperative CommunicationCooperative communications have been extensively studied in the literature.The basic idea behind cooperative communication is as follows: when the sourcetransmits messagetothedestination, thenodesinbetweencanalsoreceiveitduetothebroadcastnatureofthewirelessmedia. Thosenodescanprocessthereceivedsignal andretransmit tothedestination. Therefore, thedestinationcanmake use of the multiple copies of the message to create spatial diversity to improvethereception performance. It isrecognized that cooperative communications canimprove the transmission rate, save energy, enhance the reliability and so on. In thefollowing, we only briey introduce two cooperative communication approaches:Amplify-and-Forward (AF) and Decode-and-Forward (DF).ForAFrelayingmode, thesourcersttransmits thesignaltothedestination,which is also overheard by the relay. When the relay receives the signal from thesource, it just scales the received signal by a factor, and then forwards the ampliedversion tothedestination. Afterreceiving thosecopies, thedestination combinesthemusingmaximumratiocombining(MRC)toachievetheoptimalreception.Theoverallsignal-to-noiseratio(SNR)atthedestinationisequaltothesumofthe received SNRs from both links. The advantage of AF mode is that it is simple toimplement. However, the disadvantage is that the noise at the relay is also ampliedand forwarded to the destination.For DF relaying mode, different from the AF mode, the relay rst decodes thereceivedsignal, re-encodes it, andthenforwards tothedestination.Ifthesignalisdecoded correctly, the noisecomponent from thesource node canberemoved1.4 Summary 13PU1PU2SUFig. 1.7 Cooperativecognitive radio networkingat therelaynodebeforeforwardingtothedestination. Ifthesignal isdecodedincorrectly,thentherelayedsignal ismeaninglesstothedestination. Therefore,the overall performance of DF mode is determined by the worst link between thelink from the source to the relay and the one from the source to the destination. Inaddition, to further improve the performance, adaptive DF mode can be employed,whichallowstherelaytoforwardtheinformationonlywhentherelaydecodessuccessfully. Compared with AF mode, DF mode has the advantage of reducing theadverse effects of noise at the relay. However, it is more complex.1.3.2.2 CCRNBecauseof thebenetsof cooperativenetworking, thereisastronginterest tointroducecooperativenetworkingtoCRNstodeal withchallengesofspectrumsensingandbetter exploretransmissionopportunities. As showninFig. 1.7, incooperative cognitive radio networking (CCRN), SUs cooperate with PUs toimprove thelattersperformance interms oftransmission rate,reliability, energyefciency and so on, and in return gain transmission opportunities [35, 36, 4854].Specically,anSUactsasarelaytoimproveaPUstransmissionperformance.Then, the PU grants a period of time to the SU as a reward, in which the SU canaccessthespectrumbands fortransmissions.Byleveraging cooperation betweenPUsandSUs, awin-winsituationiscreated, wherethePUsperformanceisenhanced and SUs can access the channel in the rewarding time. By this emergingcooperative networking, SUs can be relieved from the burden of spectrum sensing.The details regarding CCRN are presented in Chap. 2.1.4 SummaryIn this chapter, cognitive radio and cognitive radio networks are introduced, whichare envisioned to increase spectrum utilization and solve the problem of spectrumscarcity.Spectrumsensing, asakeycomponent ofcognitiveradio, isdiscussed,and the limitations are highlighted. Two forms of cooperation, cooperative sensingand cooperative cognitive radio networking, areintroduced tosolve theissues ofspectrum sensing. In this book, we mainly focus on the latter form of cooperation.Chapter 2Cooperative Cognitive Radio NetworkingAbstract Asapromisingparadigm,cooperativecognitiveradionetworkinghasgainedprominent attentionsintheliterature, wherebycooperativenetworkingisleveraged tocreateawin-winsituationforbothPUsandSUs. Thischapterwillrst providea comprehensivesurveyof existingliteratureinCCRNtobetterunderstand the issues related. Then, the security aspects are discussed, which areof great importance and need to be considered before the widespread deployment ofcooperation in CRNs.2.1 Literature SurveyThissectionwill surveythestate-of-the-artresearchinCCRNinthefollowingcategories: (1) networkarchitecture; (2)cooperationsetting; (3)cooperationondifferent layers; (4) communication scenario; (5) cooperation phase and (6) cooper-ation objective. In each category, the issues and challenges are provided, followedby a discussion on future work.2.1.1 Network ArchitectureAsmentionedbefore, CRNsarecomprisedof twocomponents: aprimarynet-workandasecondarynetwork, either of whichcanbeinanadhocmodeoraninfrastructuredmode. Cooperationfor different networkarchitectures posesdifferent challenges and have different features. Cooperation between twoinfras-tructured networks is studied in [55, 56]. In [55], cooperation is studied under thescenarioconsistingofaninfrastructuredprimarynetworkandaninfrastructuredsecondarynetwork, wherethePUscommunicatewiththeprimarybasestationandSUscommunicatewiththesecondarybasestation. Theobjectiveof coop-erationisformulatedasaweightedsumthroughput maximizationproblem,andN. Zhang and J.W. Mark, Security-aware Cooperation in Cognitive Radio Networks,SpringerBriefs in Computer Science, DOI 10.1007/978-1-4939-0413-6__2, The Author(s) 20141516 2 Cooperative Cognitive Radio Networkingclosed-form solutions of the optimal power setting/allocation are obtained for theamplify-and-forward and decode-and-forward relaying modes, respectively. In [56],theauthors takeintoaccount that bothactiveandinactivePUs coexist intheprimary network, and propose two simple cooperation frameworks that are capableof stimulating both active and inactive PUs, and SUs to participate in cooperationto achieve mutual benets. Specically, for active PUs, SUs can relay PUs packetsand obtains transmission opportunities as a reward. For inactive PUs, neighboringSUs can lease spectral bands together from inactive PUs, and perform cooperativecommunication with each other. In [57], the cooperation is considered for a scenario,where the primary network is an infrastructured network and the secondary networkis an ad hoc network. Multiple SUs compete with each other to maximize their ownutilities via cooperation with the PU and the PU selects the best SU with which itcan achieve the maximum utility.2.1.2 Cooperation SettingThe cooperation settings under consideration includes: (1) single PU and single SU;(2) single PU and multiple SUs; and (3) multiple PUs and multiple SUs. For single-usercooperation, themainissueisresourceallocation, suchaspowerandtimeallocation. For multi-user cooperation, the issues of relay selection and coordinationneed to be considered.Cooperationschemes betweenonesingle PUandonesingle SUbasedoncooperative amplify-and-forward (AF) and decode-and-forward (DF) relaying areproposed respectively in [58] and [51]. In [58], AF cooperative communication isemployed when the SU acts as a helper to relay the PUs signal, while in [51], theSU rst decodes PUs signal and then do the forwarding to improve the transmissionperformance of the PU. In [59], a more general multi-user scenario is considered,where there exists one primary pair (e.g., a primary transmitter and receiver) andmultipleSUs seekingtransmissionopportunities. Adistributedsecondaryuserselection scheme is proposed, which optimizes the performance of the secondarysystem without degrading the performance of the primary system. Specically, theSU, which can help the PU to achieve the targeted transmission rate, is selected torelay the PUs message. In the meanwhile, the SU which can minimize the outageprobability of the secondary network is selected to transmit simultaneously, underthe constraint that its interference should not be above the interference threshold ofthe PU.In[35, 36],thecooperation betweenasinglePUandmultipleSUsisstudiedbyusingStackelberggame, wherebythePUcooperateswithaset of SUs toincrease PUs transmission rate [36], or to improve the PUs utility in terms of thetransmission rate and the revenue obtained from SUs [35]. In the rewarding time,multiple SUs share the spectrum granted by the PU, either using a power allocationgame [36] or a payment mechanism [35]. A similar scenario can be found in [54],where multiple SUs compete to cooperate with the PU by acting as a relaying node2.1 Literature Survey 17to gain the transmission opportunities. The relaying SU is selected among multiplecandidates using an auction mechanism, where the PU, the competing SUs and theaccess time slot are modeled as the auctioneer, the bidders and the bidding article,respectively. In [49], the authors extend the framework of CCRN by considering thepresence of multiple PUs. Multiple PUs compete with each other for cooperationwith SUs which have QoS requirements on spectral resources received for their owntransmission.Thescenarioismodeled asageneralized Nashequilibrium (GNE)problem, and the GNE and variational inequality solutions are discussed.Inaddition, the authors in[37, 52] consider the existenceof multiplePUsperformingcooperationwithmultipleSUsinthenetwork. Specically, in[52],the transmission of PUs are divided into different frames and different pairs of PUand SU perform cooperation over different frames to maximize the network utility.In [37], cooperation for multi-channel CRNsisinvestigated, where multiple PUsoperating over different channels cooperate withdifferent SUssimultaneously tomaximize the network utility. To this end, the maximum weight matching is utilizedto coordinate the cooperation between multiple PUs and multiple SUs.2.1.3 Cooperation on Different LayersThe cooperation between PUs and SUs can be performed on different layers, suchas physical layer, link layer and network layer. For physical layer, the cooperationobjective involves the outage probability, transmission rate, and so on. Moreover,diverse physical layertechniques canbe leveraged tofacilitatecooperation, suchas superposition coding, network coding, quadrature signalling. For the link layer,thecooperation takesadvantages oftheretransmission scheme(ARQ)toexploitcooperation benets. For the network layer, cooperation is carried out to nd theoptimal routing and so on.For cooperationonphysical layer, superpositioncodingandnetworkcodingareconsideredin[60], wheretwoPUscommunicatewitheachother withthehelp ofanSU.TheSUrstdecodes thesignals from thetwoPUs,then employnetwork coding to encode them. After that, the SU superposes its own signal withthe network-coded primary signals and broadcast using different power levels. Theoutage probabilities for both the primary system and the secondary system underthe proposed cooperation scheme are derived. Moreover, the cooperation range isobtained, and once the SU is located in that range, the cooperation is benecial toboth the primary system and secondary system.In [61, 62], quadrature signaling is utilized during cooperation between PUs andSUs. In[61], acooperationschemebasedonquadraturesignalingisproposed,wherebythePUandthecooperatingSUusequadratureamplitudemodulation(QAM) toattainorthogonal signalingtocooperateefciently. TheSUselectstheoptimal powerallocationcoefcient tomaximizetheperformance ofthePUwhenitsowntransmissionrequirement issatised. In[62],quadrature signalingis utilizedfor SUs tosharethe spectrumreceivedfromthePU. In[50], the18 2 Cooperative Cognitive Radio Networkingcapabilityofmultipleantennasisexploitedtofacilitatecooperation,whereSUsareequippedwithmultipleantennasandPUsareequippedwithsingleantenna.When PUs are transmitting, SUs can receive signals from another SU at the sametime if the total number of on-going trafc stream is no more than the Degree-of-Freedom (DoF) provided by SUs antennas. When SUs are forwarding PUs signal,SUs can simultaneously transmit their own signals using beamforming to mitigateinterference to the primary receivers.Cooperation on link layer is investigated in [54, 6365], which take advantage ofopportunities thatariseduringAutomaticRepeatreQuest(ARQ)retransmission.In[54], cooperativeARQisintegratedintothecooperationscheme, wherebyarelaying SU is employed if needed and it is also rewarded to use a part of the timeslot for its own transmission. The relaying SU is selected from multiple candidatesusinganauctiongame. Acooperationschemeis proposedin[63] for theSUtoexploittransmissionopportunitiesintheARQbasedprimarysystemwithoutdegradingthePUsperformance. Specically, therearetwomodesfortheSU:cooperation mode and access mode. For cooperation mode, the SU acts as a relay tohelp the PUs transmission and accumulate credits for use in the future. For accessmode, the SU transmits simultaneously when the PU is retransmitting its message.To make sure that the primary system achieves higher throughput on average, thecreditsobtainedfromthecooperationmodeshouldbesufcient tocompensatetheperformance degradation oftheprimarysysteminaccessmode.In[65], theSU overhears the ACK/NACK feedback sent from the primary receiver, and thendecides to access the spectrum or not. An opportunistic sharing scheme is proposedto exploit four kinds of spectrum opportunities based on the ACK/NACK of PUs.Cooperationonnetworklayerisstudiedin[66], wherethereexistaprimarymulti-hop network and a set of SUs. Opportunistic routing is employed to improvethe throughput of the primary network over fading channels, which selects the nexthop in anopportunistic wayaccording tothedecoding outcomes of theprevioustransmission. To exchange for spectrum access opportunities, the SUs can serve aspotential next hops that route packets based on opportunistic routing.2.1.4 Cooperation PhasesIn terms of cooperation phases, all existing work can be divided into two groups:two-phase cooperation schemes and three-phase cooperation schemes. For the three-phase cooperation, the PU transmits in the rst phase, then the SUs relay the PUssignal inthesecond phase, and inthelastphase theSUsaccessthespectrum asareward.Intheliterature,thereareampleworksthatmakethistypeofschememore efcient and realistic. On the other hand, different physical layer techniques,e.g., quadrature signalling, beamforming, superposition coding, network coding, areleveraged to merge the last two phases to facilitate two-phase cooperation schemes.In [36], a three-phase cooperation scheme is proposed, whereby the PU transmitsmessagetoaset ofSUsintherst phase, theSUswhichdecodethemessage2.1 Literature Survey 19successfully relay the PUs message via distributed space-time coding (DSTC). Inthelast phase, thecooperating SUsstart their transmissions by selecting suitablepower levels. In [35], the cooperation between PUs and SUs is also performed in athree-phase fashion, whereby SUs cooperate with the PU to improve the PUs utilityand then share the rewarding resource via a payment mechanism.A two-phase cooperation scheme is proposed in [51], whereby the PU transmitsits signal to the SU in the rst phase, and then the SU decodes the received signalandsuperimposesitwithitsownsignaltobroadcastinthesecondphase, usingdifferent power levels. In [55], the authors propose a two-phase cooperation scheme,wherebythePUselectsanSUforcooperation tomaximizethethroughput, andthe selected SU relays the PUs signal and transmits its own signal simultaneouslybyleveragingthe degrees of freedomprovidedbyorthogonal modulation. In[67], a novel polarization enabled two-phase cooperation framework for cognitiveradio networking is proposed. By leveraging the degrees of freedom provided byorthogonally dual-polarized antennas, secondary users can relay the trafc of PUsand transmit their own trafc in the same time slot without interference.In[58], atwo-phaseschemebasedoncooperativeAFrelayingprotocol isproposed. The key feature of the proposed approach is that the SU linearly combinesthe primary signal with the secondary signal and allocates fractions z and 1 z ofthe transmission power to the primary and secondary signals separately. It is shownthat for a xed z, a critical region can be found where the outage probability of theprimary system is kept lower than the case without spectrum sharing.In[68], networkandsuperpositioncodingareutilizedtofacilitatetwo-phasecooperation for the case where multiple primary channels exist. Two PUs transmitdatatoacommondestinationwiththeassistanceofanSUthat actsasarelayand the SU attains the opportunity of transmission in return. Network coding andsuperposition coding are utilized at the relay SU to superimpose its own messageon network-coded primary signals. The probabilities of outage regions are obtainedfor multiple access channels.2.1.5 Cooperation ObjectiveThe cooperation objective for PUs and SUs could be the same or different, whichhasdifferent approaches tomodel theinteractionsbetweenPUsandSUs. Intheliterature, the objectives of PUs include transmission rate, reliability, energy saving,security, connectivity, and so on.In [35, 36, 49, 53], the objective of cooperation is to maximize the PUs trans-mission rate and provide transmission opportunities to SUs. In [36], the cooperatingSUs relay the PUs codeword via DSTC and share the rewarding resource using apower allocation game. In [35],SUscooperate withthePUtoimprove thePUsutility, which is a combination of transmission rate and the revenue obtained fromSUs. Then, multiple cooperating SUs share the rewarding resource via a paymentmechanism.20 2 Cooperative Cognitive Radio NetworkingIn [54, 66], cooperation is performed by considering the reliability of the PU. In[54], a novel distributed scheme that integrates cooperative ARQ in the cooperationscheme is proposed. The SU which successfully decodes the PUs message can actas a relay to increase the reliability of primary transmission, and in return gain acertain period of time for its own transmission.In [57, 69], the energy efciency of the primary system is considered. In [69],the authors propose a frequency-division multiple access (FDMA) based two-phasecooperation in which a PU divides its spectrum into two orthogonal subbands andbroadcasts on the rst subband in the rst phase. SUs relay the PUs signal on thesamesubbandinthesecondphase,andcontinuously transmitinbothphasesonthe second subband. In [57], a cooperation scheme based on time-division multipleaccess(TDMA)isproposed, whereanSUcooperates withaPUtoimprove theenergy saving of the primary transmission and gains transmission opportunities. Asthe license holder, the PU can decide when to cooperate, with whom to cooperate,and how to cooperate.In [7072], cooperation schemes are studied, which improve the physical layersecurityoftheprimarylinkandprovidetransmissionopportunities toSUs. Thesecuritylevel is measuredbysecrecyrate, whichis denedbythedifferencebetween the transmission rate at the primary receiver and that at the eavesdropper.The objective of cooperationis to maximize the secrecyrate of the primarytransmission.2.2 Security AspectsCooperationcanbringmanybenetsifallnodesarewell-behaved. However,inreality, this assumption may not hold. When there exist some dishonest or maliciousSUs, cooperation can incur security issues, which will compromise the cooperationbenets anddisturbthenormal operationof CCRN. Ontheother hand, sincecooperation can be leveraged to increase the secrecy of transmission, security alsobrings opportunities for cooperation.2.2.1 Trust-Aware CooperationInCCRN, thePUspacketscanbeeavesdroppedbytheneighboringSUssuchthat the condentialitycannot beguaranteed. Moreover, the malicious relayscanalterthePUspacketsorfabricateitspacketsandthenforwardthemtothedestination. TosecurecooperationbetweenPUs andSUs, thefollowingbasicsecurity requirements need to be met: condentiality, integrity, and authentication,which can be provided by suitable cryptographic approaches (e.g., encryption anddecryption, authentication, messageauthenticationcode, digital signature, etc.).However, a legitimate SU may be compromised and misbehaves during cooperation2.2 Security Aspects 21whenit isselectedtocooperatewiththePU, e.g., it maylaunchblackorgreyholeattack[73]. AdishonestSUmaynot obeythecooperationruleduringthecooperative transmission to pursue more self-benets, e.g., it may transmit its ownpackets instead of relaying the PUs packets. Furthermore, considering the mobilityof SUs, the malicious or dishonest SUs may misbehave at one place then move toother places. Since thereis no record ofthe past behaviors, theseusers canhavethe same opportunity to participate in cooperation with the PU, and then continueto harm the system. In a nutshell, without considering these security threats, the PUmay choose an untrustworthy SU for cooperation, which will cause the failure ofthe cooperation and degrade the PUs quality of service (QoS).As a summary, the potential misbehaviors in CCRN can be listed as follows.1. Selshness:thecooperating SUmaychoosealowertransmissionpowerthanthe expected one during cooperation or it just chooses not to forward the PUsmessage to save energy.2. Maliciousness: the malicious SU may delete, modify or replace the bits in the DFmode. In AF mode, it may intentionally add some jamming signals to corrupt thePUs signal.3. Dishonesty:thedishonestSUmaypresentthefakeCSItogaintransmissionopportunities.Therefore, the security issues need to be considered, when PUs choose to cooperatewith SUs.2.2.2 Cooperation for SecrecyAsmentioned above, cooperation canincurdiverse securityissues.Inthemean-while, security also brings opportunities for cooperation. Specically, in a hostileenvironment, thereexist someunfriendlyusers, e.g., eavesdroppers. Duetothebroadcast nature of wireless communication, these unfriendly users can easily over-hear the ongoing transmission. This consequence not only hurts the condentialityof communications, but also exposes the vulnerabilities that a malicious user canexploit tolaunch attacks tothenetwork. Toprotect thesecrecy oftransmissions,the traditional solution is to use encryption at upper layers of the communicationprotocol. However, the security scheme at upper layers is prone to potential attacks.Moreover, it becomes very challenging for a network without infrastructure [74].In addition, in a network where the nodes have relatively low power, e.g., sensors,it mightnotbepracticaltoimplementcryptographicalgorithms[74]. Tosecurethecommunicationeffectively, thereisanovelapproachat thephysical(PHY)layer [7476], which exploits the characteristics of the wireless channel for securetransmission. In [75], it is shown that perfectly secure information can be exchangedat a nonzero rate between the source and destination. However, it becomes infeasiblewhen the source-destination channel is worse than the source-eavesdropper channel.In [77], the authors studied secure communications in the low SNR regime, which22 2 Cooperative Cognitive Radio Networkingcorresponds to the cases of long distance transmissions or energy-limited scenarios.The problemis that all the nodes are assumed to be equipped with multiple antennas,which may be infeasible inreality. To address the above issues,user cooperationhas been introduced to enhance the secrecy of communications [78]. In the contextof CCRN, a cooperation based spectrum access has been proposed in [70], whichimproves the security of the primary link and provide transmission opportunities toSUs. Nevertheless, it is achieved at the expense of employing multiple antennas aswell. Observing the above, we are inspired to design cooperation schemes for CRNs,where all users are only equipped with single antenna to enhance the security of theprimary link and provide transmission opportunities to SUs.2.3 SummaryIn this chapter, we focus on cooperative cognitive radio networking and a literaturereviewisprovided, wherecooperation schemesareclassiedintodifferent cate-gories. The security aspects of this cooperative paradigm and the limitations of theexisting works are discussed. In the subsequent chapters, security-aware cooperationschemes will be studied.Chapter 3Trust-Aware Cooperative NetworkingAbstract In this chapter, we introduce a trust-aware cooperation scheme tofacilitate the cooperation and SU selection in an unfriendly environment, wherebythe primary users (PUs) choose trustworthy partners as relays to improve throughputor energyefciency. Specically, the cooperationinvolves a PUselectingthemost suitablesecondaryuser(SU)torelayitsmessageandgivingtheselectedSUspectrumaccess right as areward, takingthetrustworthiness of SUs intoconsideration. While the SU, being starved for transmission opportunities,chooses asuitablepower level for cooperationandits owntransmission. Theproposed cooperative strategy, including partner selection, time slot allocation andpowerallocationinanunfriendlyenvironment, areanalyzed. Numerical resultsdemonstrate that, with the proposed scheme, the PU can achieve higher throughputor energy saving through cooperation with the trustworthy SU.3.1 IntroductionCooperative networking can be leveraged to deal with the limitations of spectrumsensing [31], whereby SUs cooperate with PUs to improve the PUs transmissionperformance, andinreturngaintransmissionopportunities.Therefore,bothPUsandSUscanbenet fromcooperation, whichcreatesawin-winsituation.Intheliterature,thisparadigmisreferredtoascooperativecognitiveradionetworking(CCRN) [3537, 4851, 54]. Thecommonassumptionfortheexistingworksisthat SUs are trustworthy and well-behaved during cooperation, which may not bealways true in reality. In an unfriendly environment, where there exist selsh, evenmalicious SUs, security issues mentioned in Chap. 2 can arise, which will compro-mise the normal operation of CCRN. Thus, security needs to be considered for thisemerging cooperative networking, which has not been given due consideration inthe literature.Inthischapter,wemakeaneffort toguaranteeandimprove theperformanceofcooperationandSUselectioninanunfriendlyenvironment,whereSUsmayN. Zhang and J.W. Mark, Security-aware Cooperation in Cognitive Radio Networks,SpringerBriefs in Computer Science, DOI 10.1007/978-1-4939-0413-6__3, The Author(s) 20142324 3 Trust-Aware Cooperative Networkingmisbehaveduringcooperation. Moreover, cooperationundertwoscenarioswithdifferent primary linkqualities areconsidered, wherethePUmayhavedifferentconcerns. In particular, if the primary link is poor so that the transmission qualitydrops dramatically, the PU has the incentive to increase the throughput via coopera-tion with SUs so that the QoS requirement can be satised; if the channel conditionis good enough, since the PU can meet the trafc demand on its own, it may have aconcern with energy efciency due to power limitation. By the same token, the SUmay value the throughput more since it typically does not have much transmissionopportunities. Based on the fact that the PU and the SU have different incentivesfor cooperation, weproposeacooperativespectrumaccess schemewhichalsoincorporates trust value into the system to mitigate the security issues. Consideringthat PUs and SUs aim at maximizing their own utilities, we leverage game theorytoanalyzethecooperation scheme.Furthermore, asthelicenseholders,thePUshave higher priority on spectrum usage and are supposed to lead the cooperation.Thus, we model the interaction between the PU and the SU as a Stackelberg game[79], which provides the PU with the best strategy in the partner selection process,spectrumaccess time allocation and transmission power determination. Based on theanalysis of the game, the PU selects the best SU and optimal cooperation parametersto maximize the utility if cooperation is agreed upon. That is, in accordance withtheoutcomes ofthegame, thePUcandecidewhentocooperate, withwhomtocooperate and how to cooperate.3.2 System ModelThis section presents the details of system model and the main system parameters.3.2.1 MAC LayerAsshowninFig. 3.1,thereexisttwotypesofnetworks,theinfrastructure-basedprimary network and ad hoc secondary network, collocated in the same area. In theprimary network, the PUs communicate with the base station (BS) in time-divisionmultiple access (TDMA) mode, while theSU transmits data to itscorrespondingreceiver in the secondary network. The time slot duration for each PUs transmissionisdenotedbyT . Forcooperation, thePUselectsoneSUasacooperating relayandAmplify-and-Forward (AF)protocol isemployed.ThePUgrantstheuseofthe bandwidth to the cooperating SU so as to improve the energy efciency of thecommunication to the BS. Specically, a fraction of T(0 < _ 1) is used for theprimary and cooperative transmission. The PU transmits data to the SU in the rstT2, which can be also overheard by the BS (Fig. 3.1a). In the subsequent durationofT2, theSUrelaysthereceiveddatatotheBS(Fig. 3.1b). Intheremaining3.2 System Model 25...PUPUPUSUBS(1a)T(1a)TTTTaT/2 aT/2aT/2 aT/2aT/2 aT/2aT(1a)T aTaTSUSUBSBS...... ...... ...a bcFig. 3.1 The proposed framework of cooperation in CRNs. (a) Primary transmission, (b) cooper-ative transmission, (c) secondary transmissiondurationof (1 ) T , thecooperatingSUisallowedtotransmititsowndatato the corresponding secondary receiver (Fig. 3.1c). A common control channel isconsidered to be available for exchanging information among PU, SU and BS (e.g.,channel state information (CSI), trust values, etc.) and for delivering the decision ofthe PU (e.g., slot allocation, the selected SU, etc.) to the secondary network.3.2.2 Physical LayerThe channels betweennodes canbe modeledas independent proper complexGaussianrandom variables, constant withineachslot,but generally varying overthe slots.Weusethe following notations to denote the instantaneous channels ineachslot: hbdenotesthecomplexchannel gainbetweenthePUandtheBS;26 3 Trust-Aware Cooperative Networkinghisdenotes thechannel gainbetweenthePUandSUi; hisbdenotes thechannelgain betweenSUiand the BS; andhisdenotes the channel gain betweenSUiandits corresponding receiver. The PU uses power1Jwatts for transmission withoutcooperation. AsshowninFig. 3.1, withcooperation, thePUchoosespower 1cwattsforprimary transmission.Thebandwidth ownedbytheprimary userisWHz.SUiis constrained to spend the same power1isfor both the cooperative andsecondary transmission so as to ensure SUispend at least the same power which itis willing to spend for its own transmission. The one-sided power spectral density ofthe independent additive white Gaussian noise at the both base station and secondaryreceivers is N0.3.3 Cooperation for Throughput ImprovementIn this section, we will discuss the cooperation when the primary link is weak. Dueto the poor channel condition, the signal reception quality at the destination receiverwill be degraded dramatically. In order to meet the trafc demand, the PU selectsa suitable relay to improve the throughput. The objective of the PU is to maximizethe throughput through cooperation in an unfriendly environment. To evaluate therisks of cooperation, trust value is applied and the cooperation procedure is modeledusing Stackelberg game. In such a game, the utilities of both the PU and the SU arepresented and analyzed; and the close-form solutions for the players best strategiesare derived, which constitute the Stackelberg equilibrium.3.3.1 Trust Computational ModelIn anunfriendly environment, the aforementioned security issuescan rise,whichcannot bewell mitigatedbymeansofcryptographicmethodologies[80]. Thus,trust andreputationsystemisappliedtoaddresstheseissues[81]. Specically,trust values are assigned to SUs and utilized to evaluate the behaviors of SUs. ThePU maintains a table for recording identities and the corresponding trust values ofits one-hop neighboring SUs. In addition, BS keeps the trust values of all SUs initsdomain. Eachtimeaftercooperation, thebehavior oftheselectedSUwillbeevaluated and the trust value will be updated accordingly. Then, the trust value willbe exchanged periodically between the PUs and the BS.We use a Bayesian framework [82, 83] to evaluate the trust values: each entity isassumed to behave well with probability , and misbehave with probability (1),i.e., the behavior of the entity follows a Bernoulli distribution. Through a series ofobservations, a posteriori probability can be derived to estimate the future behaviorsof the entity. Posteriori probabilities of binary events can be represented as the beta3.3 Cooperation for Throughput Improvement 27distribution. An expression of the probability density function (PDF)( [k. |) interms of the gamma function 1is given by:( [k. |) =1(k |)1(k)1(|) (k-1) (1 )(|-1). (3.1)where is the estimate of, andk,| are the two parameters. The expectation ofbeta distribution is given by 1( ) =k(k|), which can be used to represent the trustvalue of the relevant entity.In our system, a malicious or dishonest SUi behaves well with probability i andmisbehaves with probability 1 i. In order to estimate the trustworthiness of SUs,BSneeds toobserve theongoing transmissionandevaluate theactivitiesofSUsaccording to the received signals. To determine whether the relaying SU misbehavesornot, oneapproachistoutilizetracingsymbols,whichareknownatboththesource and the destination [84, 85]. Another way is based on the correlation betweensignals received from thesourceand therelay[86].Inaddition, themisbehaviorcanalsobedetectedbasedonthesuccessor failureof transmittedframesviaacknowledgment (ACK/NACK) [87].Based onthe related work inthe literature,the misbehavior of relaying nodes can be detected, which is beyond the scope ofthiswork. Consideraprocess withtwopossibleoutcomes (misbehavior orwell-behavior) and let j be the observed number of good behaviors and v be the observednumber of misbehaviors. Then, the PDF of observing outcomes in the future can beexpressed as a function of past observations by setting: k =j 1 and | =v 1.Thus, the expected value of can be determined from observations as follows:1( ) =j 1(j v 2). (3.2)which is used as the trust value T riof SUi.When new observations of a particular SU are made, e.g., observed misbehav-iors and observed good behaviors, the associated trust value can be updated using(3.2) by setting v := v and j := j .3.3.2 Stackelberg Game between PU and SUStackelberggame is appliedto model the cooperationprocedure, consideringdifferentprioritiesforspectrumusageoftheprimarysystemandthesecondarysystem. In the Stackelberg game, the PU acts as the leader and the SU is the follower.Astheleader, thePUcanchoosethebest strategies, awareoftheeffect ofitsdecision on the strategies of the follower (the SU); while the SU can just choose itsown strategies given the selected parameters of the PU. The utility functions for bothPU and SU are respectively dened. By analyzing the game, the best cooperatingSU and the optimal cooperation parameters can be determined.28 3 Trust-Aware Cooperative Networking3.3.2.1 Primary UserUnder the scenario of poor channel condition, the throughput of direct transmissionisreduced.Thus,thePUisinterestedincooperation toincreasethethroughput.Given a xed time duration T , increasing the throughput is equivalent to increasingthe average transmission rate. To this end, the PU selects the most suitable SU fromthe set S of its one-hop neighbors. Suppose SUiis chosen for cooperation, the PUdecides the time allocation parameter iand its transmission power 1icto maximizethroughput, on the basis of available instantaneous CSI. The average transmissionrate 1icthrough AF cooperative communication between the PU andSUiis givenas follows:1ic =iW2log2_1 1ichb2N0_1ichis2. 1ishisb2__. (3.3)where_1ichis2. 1ishisb2_=1N01ichis21ishisb21ichis21ishisb2N0.The factori2accounts for the fact that iTis used for cooperative relaying, whichis further split into two phases. Considering the trust value T riof each neighboringSUi, the utility function is given byUi = T ri1ic. (3.4)which indicates the expected throughput the PU can attain through cooperation withSUi. The objective of the PU is to maximize its utility function and the strategy isto choose the most suitable SU from the set of its one-hop neighboring SUs and thecooperation parameters, i.e., the time allocation parameters iand the transmissionpower 1icfor cooperation with the selected SUi.3.3.2.2 Secondary UserThe SUcan gain transmission opportunities through cooperation with thePU. Inparticular, the SU relays PUs data in the second phase and transmits its own datain the last phase. Assuming cooperation with the PU, the selectedSUidecides itstransmission power, pertaining tothegivenand1c.Thetarget of theSUistomaximize throughput (equivalent to thetransmission rate) without expending toomuch energy. Following the cooperation agreement,SUispends thesamepower1isfor both cooperative and secondary transmissions. In particular, the transmission3.3 Cooperation for Throughput Improvement 29rate1isfor secondary transmission (Fig. 3.1c) betweenSUiand its correspondingreceiver is given by1is(i) = (1 i)Wlog2_1 1ishis2N0_. (3.5)With energy consumption 1is(1 i2 )T , the utility function of SUican berepresented by 1is(i)T c1is(1 i2 )T , where c (0 [h11[ and [hJ1[ > [hJD[, wehave[1>0, [2>0, and11 =1Cma.. If[3>0, there is no positive root for the quadratic function in (4.13) anddd1J>0fortherangefrom0to1Cma.. Thus, 1=Jequalsto0tomaximizethesecrecy rate, indicating anon-jamming scenario. If[31Cma.,dd1J