ch11-operatorsinsharedspectrum

8/11/2019 ch11-OperatorsInSharedSpectrum

1/33

2007 Levente Buttyn and Jean-Pierre Hubaux

Security and Cooperationin Wireless Networks

http://secowinet.epfl.ch/

Chapter 11: Wireless operators in sharedspectrum

multi-domain sensornetworks;

border games incellular networks;


2/33

Security and Cooperation in Wireless NetworksChapter 11: Wireless operators in shared spectrum

2/33

Chapter outline

11.1 Multi-domain sensor networks

11.2 Border games in cellular networks


3/33


3/33

Typical cooperation: help in packet forwarding

Can cooperation emerge spontaneously in multi-domainsensor networks based solely on the self-interest of thesensor operators?

Multi-domain sensor networks



4/33


4/33

Simplified model

C: Cooperation D: Defection

4 possible moves:

CCthe sensor asks for help (cost 1) and helps if asked (cost 1) CDthe sensor asks for help (cost 1) and does not help (cost 0)

DCthe sensor sends directly (cost 2a) and helps if asked (cost 1)

DDthe sensor sends directly (cost 2a) and does not help (cost 0)

2

11

1 1 1

11.1 Multi-domain sensor networks11.1.1 Simplified model


5/33


5/33

Example : CCCD (1/6)

CCthe sensor tries to get help from the other sensorand helps if the other sensor requests it

CDthe sensor tries to get help but it refuses to help

CC CD



6/33


6/33




CC CD



7/33


7/33


CC

failure

CD





8/33


8/33


CC CD





9/33


9/33


CC CDsuccess



10/33


10/33


CC CD

Black player

Cost: 2

1 for asking

1 for helping

Benefit: 0

(packet dropped)

Gray player

Cost: 1

1 for asking

Benefit: 1

(packet arrived)



11/33


11/33

2

11

Cost for blackCost for grey

Outcome for black (0 = failure)

Outcome for grey (1 = success)

The simplified model in strategic form



12/33


12/33

Reception threshold

time

success / failure of packet reception

Sliding window of history

Success

(= 1)

Failure

(= 0)

Reception

threshold

Average of the packet reception

Risk of going below threshold

adapt strategy (move to theconstrained state: only DC or DD

are eligible)

Reception threshold: computed and stored at each sensor node

The battery (B) level of the sensors decreases with the moves

If the battery is empty, the sensor dies



13/33


13/33

Game Theoretic Approach

The mentioned concepts describe a game

Players: network operators

Moves (unconstrained state): CC, CD, DC, DD

Moves (constrained state): DC, DD

Information sets: histories

Strategy: function that assigns a move to every possiblehistory considering the weight threshold

Payoff = lifetime

We are searching for Nash equilibria with the highest

lifetimes



14/33


14/33

Two-step Strategies

Binitial battery

reception threshold

apath loss exponent (2)

1,2payoff of transient states

Cooperative Nash equilibrium

Non-cooperative Nash equilibrium

If > 1/3, then (CC/DD, CC/DD) is more desirable



15/33


15/33

Generalized Model

11.1 Multi-domain sensor networks11.1.2 Generalized model

Simplified model with the following extensions: many sensors, random placing

minimum energy path routing common sink / separate sink scenarios

classification of equilibria Class 0: no cooperation (no packet is relayed)

Class 1: semi cooperation (some packets are relayed)

Class 2: full cooperation (all packets are relayed)


16/33


16/33

Main simulation parameters

Parameter Value

Number of sensors per domain 20

Area size 100 x 100 m

Reception threshold 0.6

History length 5

Path loss exponent 234 (3)



17/33


17/33

Impact of the path loss exponent

Percentageofs

imulations

Equilibrium classes ( 0no cooperation, 1semicooperation, 2full cooperation)

Value of the pathloss exponent

2

3

4



18/33


18/33

Conclusion on multi-domain sensor networks

We examined whether cooperation is possible without theusage of incentives in multi-domain sensor networks

In the simplified model, the best Nash equilibria consist ofcooperative strategies

In the generalized model, the best Nash equilibria belong tothe cooperative classes in most of the cases



19/33


19/33

Chapter outline




20/33


20/33

Motivating example



21/33


21/33

Introduction

spectrum licenses do notregulate access overnational borders

adjust pilot power toattract more users

Is there an incentive for operators to apply competitive

pilot power control?



22/33


22/33

System model (1/2)

Network:

cellular networks using CDMA channels defined by orthogonal

codes

two operators: A and B

one base station each

pilot signalpower control

Users:

roaming users

users uniformly distributed

select the best quality BS

selection based signal-to-interference-plus-noise ratio(SINR)

11.2 Border games in cellular networks11.2.1 Model


23/33


23/33

System model (2/2)

0

pilot

p i ivpilotiv pilot pilot

own other

G P gSINRN I I

W

i

pilot

own iv iw

w

I g T

M

i

pilot

other jv j iw

j i w

I g P T

M

A Bv

PA

PB

TAv

TBw

TAw

0

tr

p iv ivtr

iv tr tr

own other

G T gSINR

N I I

W

, i

pilot

own iv i iw

w v w

I g P T

M

tr pilot

other other I I

pilot signal SINR:

traffic signal SINR:

Pi

pilot power of i

processing gain for the pilot signal

pilot

pG

ivg

0N noise energy per symbol

W

ivT

pilot

ownI

channel gain between BS iand user v

available bandwidth

own-cell interference affecting the pilot signal

own-cell interference factor

traffic power between BS iand user v

other-to-own-cell interference factor

iM set of users attached to BS i

11.2 Border games in cellular networks11.2.1 Model

h d l


24/33


24/33

Game-theoretic model

Power Control Game, GPC players networks operators (BSs), A and B strategy pilot signal power, 0W < Pi < 10W, i= {A, B}

standard power, PS= 2W

payoff profit, where is the expected income

serving user v normalized payoff difference:

i

i v

v

u

M

v

max , ,

,

i

S S S

i i is

iS S

i

u s P u P P

u P P

11.2 Border games in cellular networks11.2.2 Power control game

Si l i i


25/33


25/33

Simulation settings


I th ?


26/33


26/33

Is there a game?

only A is strategic (B uses PB= PS)

10 data users

path loss exponent, = 2

i


Wh b th t t t i


27/33


27/33

When both operators are strategic

10 data users

path loss exponent, = 4


N h ilib i


28/33


28/33

Nash equilibria

10 data users 100 data users


Effi i (1/2)


29/33


29/33

Efficiency (1/2)

10 data users


Effi i (2/2)


30/33


30/33

Efficiency (2/2)

100 data users


C t NE (1/2)


31/33


31/33

convergence based on better-response dynamics

convergence step: 2 W

Convergence to NE (1/2)

PA= 6.5 W

11.2 Border games in cellular networks11.2.3 Convergence to a Nash Equilibrium

Con e gence to NE (2/2)


32/33


32/33

Convergence to NE (2/2)

convergence step: 0.1 W

11.2 Border games in cellular networks11.2.3 Convergence to a Nash Equilibrium

Conclusion on border games


33/33

Security and Cooperation in Wireless Networks

Conclusion on border games

not only individual nodes may exhibit selfish behavior, butoperators can be selfish too

example: adjusting pilot power to attract more users atnational borders

the problem can be modeled as a game between theoperators

the game has an efficient Nash equilibrium

theres a simple convergence algorithm that drives the system intothe Nash equilibrium

ch11-operatorsinsharedspectrum

Documents