Mechanisms and Games for Dynamic Spectrum Allocation

604 Pages·2014·4.44 MB·English

by  Alpcan T.

MechanismsandGamesforDynamicSpectrumAllocation
Presentingstate-of-the-artresearchintomethodsofwirelessspectrumallocationbased
on game theory and mechanism design, this innovative and comprehensive book pro-
vides a strong foundation for the design of future wireless mechanisms and spectrum
markets.
Prominent researchers showcase a diverse range of novel insights and approaches
to the increasing demand for limited spectrum resources, with a consistent emphasis
ontheoreticalmethods,analyticalresults,andpracticalexamples.Coveringfundamen-
tal underlying principles, licensed spectrum sharing, opportunistic spectrum sharing,
andwidertechnicalandeconomicconsiderations,thissingularbookwillbeofinterest
to academic and industrial researchers, wireless industry practitioners, and regulators
interestedinthefoundationsofcutting-edgespectrummanagement
TANSUALPCANisaSeniorLecturerintheDepartmentofElectricalandElectronicEngi-
neeringattheUniversityofMelbourne,andco-authorofNetworkSecurity:ADecision
andGame-TheoreticApproach(2011).
HOLGER BOCHE is a Professor in the Institute of Theoretical Information Technology,
TechnischeUniversitätMünchen,andaFellowoftheIEEE.
MICHAELL.HONIGisaProfessorintheDepartmentofElectricalEngineeringandCom-
puterScienceatNorthwesternUniversity.HeisaFellowoftheIEEE.
H.VINCENTPOORistheDeanofEngineeringandAppliedScienceandMichaelHenry
Strater University Professor of Electrical Engineering at Princeton University. He is a
co-author of Principles of Cognitive Radio (2012), and a Fellow of the IET and the
IEEE.

Mechanisms and Games for
Dynamic Spectrum Allocation
TANSU ALPCAN
UniversityofMelbourne
HOLGER BOCHE
TechnischeUniversitätMünchen
MICHAEL L.HONIG
NorthwesternUniversity
H.VINCENT POOR
PrincetonUniversity
UniversityPrintingHouse,CambridgeCB28BS,UnitedKingdom
PublishedintheUnitedStatesofAmericabyCambridgeUniversityPress,NewYork.
CambridgeUniversityPressispartoftheUniversityofCambridge.
ItfurtherstheUniversity’smissionbydisseminatingknowledgeinthepursuitof
education,learning,andresearchatthehighestinternationallevelsofexcellence.
www.cambridge.org
Informationonthistitle:www.cambridge.org/9781107034129
(cid:13)c CambridgeUniversityPress2014
Thispublicationisincopyright.Subjecttostatutoryexception
andtotheprovisionsofrelevantcollectivelicensingagreements,
noreproductionofanypartmaytakeplacewithoutthewritten
permissionofCambridgeUniversityPress.
Firstpublished2014
PrintedintheUnitedKingdombyTJInternationalLtd.PadstowCornwall
AcataloguerecordforthispublicationisavailablefromtheBritishLibrary
LibraryofCongressCataloging-in-PublicationData
Mechanismsandgamesfordynamicspectrumallocation/[compiledby]TansuAlpcan,
UniversityofMelbourne,HolgerBoche,TechnischeUniversitätMünchen,
MichaelL.Honig,NorthwesternUniversity,H.VincentPoor,PrincetonUniversity
pagescm
Includesbibliographicalreferencesandindex.
ISBN978-1-107-03412-9(hardback)
1. Wirelesscommunicationsystems–Management.2.Radiofrequency
allocation.3.Signaltheory(Telecommunication)4.Gametheory.I.
Alpcan,Tansu,1975-editorofcompilation.
TK5103.2.M437 2013
(cid:48)
384.54 524015193–dc23
2013044140
CambridgeUniversityPresshasnoresponsibilityforthepersistenceoraccuracyof
URLsforexternalorthird-partyinternetwebsitesreferredtointhispublication,
anddoesnotguaranteethatanycontentonsuchwebsitesis,orwillremain,
accurateorappropriate.
Contents
Contributors pagexvii
Preface xxi
PartITheoreticalFundamentals 1
1 Gamesandmechanismsfornetworkedsystems:incentivesandalgorithms 3
AnilKumarChorppath,TansuAlpcan,andHolgerBoche
1.1 Introduction 3
1.2 Systemmodel 6
1.3 Interferenceandutilityfunctionmodels 8
1.4 Pricingmechanismsformulti-carrierwirelesssystems 10
1.4.1 Netutilitymaximization 12
1.4.2 Alternativedesignerobjectives 15
1.5 Learninginpricingmechanisms 17
1.6 Auction-basedmechanisms 19
1.7 Discussionandopenproblems 28
References 28
2 CompetitioninwirelesssystemsviaBayesianinterferencegames 32
SachinAdlakha,RameshJohari,andAndreaGoldsmith
2.1 Introduction 32
2.2 StaticGaussianinterferencegames 35
2.2.1 Preliminaries 35
2.2.2 TheGaussianinterferencegamewithunknownchannelgains 36
2.2.3 BayesianGaussianinterferencegame 38
2.3 Sequentialinterferencegameswithincompleteinformation 41
2.3.1 Atwo-stagesequentialgame 41
2.3.2 Asequentialgamewithentry 44
2.4 Repeatedgameswithentry:thereputationeffect 45
2.4.1 ArepeatedSBGI-Egame 46
2.4.2 Sequentialequilibriumoftherepeatedgame 47
2.5 Conclusion 49
vi Contents
2.6 Appendix 50
References 55
3 Reactingtotheinterferencefield 57
MérouaneDebbahandHamidouTembine
3.1 Introduction 57
3.1.1 Spectrumaccessasagame 57
3.1.2 Cognitiveaccessgame 57
3.1.3 Mean-fieldgameapproach 58
3.1.4 Interferencemanagementinlarge-scalenetworks 58
3.1.5 Objectives 59
3.1.6 Structureofthechapter 59
3.1.7 Notations 60
3.2 Wirelessmodel 60
3.2.1 Channelmodel 61
3.2.2 Mobilitymodel 62
3.2.3 Path-lossmodel 62
3.2.4 Remainingenergydynamics 63
3.2.5 Queuedynamics 63
3.2.6 SINRmodel 63
3.3 Game-theoreticformulations 64
3.4 Reactiontotheinterferencefield 64
3.4.1 Introductiontomean-fieldgames 64
3.4.2 Theinterferencefield 67
3.5 Mean-fieldstochasticgame 67
3.5.1 Onagamewithone-and-halfplayer 68
3.5.2 Strategiesandpayoffs 68
3.5.3 Mean-fieldequilibrium 69
3.5.4 Structureoftheoptimalstrategy 69
3.5.5 Performance 70
3.5.6 Mean-fielddeterministicgame 70
3.5.7 Hierarchicalmean-fieldgame 71
3.6 Discussions 71
3.7 Conclusions 72
3.8 Openissues 72
Acknowledgements 73
References 73
4 Walrasianmodelforresourceallocationandtransceiverdesignin 75
interferencenetworks
EduardA.JorswieckandRamiMochaourab
4.1 Consumertheory 76
Contents vii
4.1.1 Standardconsumertheory 77
4.1.2 Consumertheoryforutilityα−βx +γx x 79
1 1 2
4.1.3 Example1:Protectedandsharedbands 80
4.1.4 Example2:Two-userMISOinterferencechannel 85
4.1.5 Example3:Multi-carrierinterferencechannel 89
4.1.6 Discussionandcomparisonofconsumermodels 91
4.2 Walrasianmarketmodel 92
4.2.1 ExistenceofaWalrasianequilibrium 92
4.2.2 UniquenessoftheWalrasianequilibrium 94
4.2.3 Convergenceofatâtonnementprocess 95
4.2.4 EfficiencyofaWalrasianequilibrium 95
4.2.5 Example1:Two-userprotectedandsharedbands 96
4.2.6 Example2:Two-userMISOinterferencechannel 99
4.2.7 Example3:MCinterferencechannel 103
References 106
5 Powerallocationandspectrumsharinginwirelessnetworks:an 108
implementationtheoryapproach
AliKakhbod,AshutoshNayyar,ShrutivandanaSharma,andDemosthenisTeneketzis
5.1 Introduction 108
5.1.1 Chapterorganization 109
5.2 Whatisimplementationtheory? 109
5.2.1 Gameforms/mechanisms 110
5.2.2 Implementationindifferenttypesofequilibria 111
5.2.3 Desirablepropertiesofgameforms 114
5.2.4 Keyresultsonimplementationtheory 115
5.3 NashimplementationforsocialwelfaremaximizationandweakPareto 118
optimality
5.3.1 Themodel(M ) 118
PSA
5.3.2 Thepowerallocationandspectrumsharingproblem 121
5.3.3 Constructingagameformforthedecentralizedpowerand 122
spectrumallocationproblem
5.3.4 Socialwelfaremaximizingpowerallocationinasingle 125
frequencyband
5.3.5 WeaklyParetooptimalpowerandspectrumallocation 127
5.3.6 InterpretingNashequilibrium 129
5.3.7 Otherapproachestopowerallocationandspectrumsharing 130
5.4 Revenuemaximization 131
5.4.1 Themodel 132
5.4.2 Impossibilityresultfromimplementationtheory 133
5.4.3 Purelyspectrumallocationproblem 133
5.4.4 Purelypowerallocationproblem 140
5.4.5 Othermodelsandapproachesonrevenuemaximization 140
viii Contents
5.5 Conclusionandreflections 141
References 142
6 Performanceandconvergenceofmulti-useronlinelearning 145
CemTekinandMingyanLiu
6.1 Introduction 145
6.2 Relatedwork 146
6.3 Problemformulationandpreliminaries 149
6.3.1 Factorsdeterminingthechannelquality/reward 149
6.3.2 Channelmodels 150
6.3.3 Thesetofoptimalallocations 151
6.3.4 Performancemeasure 153
6.3.5 Degreeofdecentralization 153
6.4 Mainresults 154
6.5 Achievableperformancewithnofeedbackandiidchannels 155
6.6 Achievableperformancewithpartialfeedbackandiidchannels 160
6.7 Achievableperformancewithpartialfeedbackandsynchronizationfor 167
iidandMarkovianchannels
6.7.1 AnalysisoftheregretofDLOE 170
6.7.2 Regretanalysisforiidchannels 172
6.7.3 RegretanalysisforMarkovianchannels 175
6.8 Discussion 179
6.8.1 Strategicconsiderations 179
6.8.2 Multipleoptimalallocations 180
6.8.3 Unknownsuboptimalitygap 182
Acknowledgements 183
References 183
7 Game-theoreticsolutionconceptsandlearningalgorithms 185
SamirM.PerlazaandSamsonLasaulce
7.1 Introduction 185
7.2 Ageneraldynamicspectrumaccessgame 186
7.3 Solutionsconceptsanddynamicspectrumaccess 187
7.3.1 Nashequilibrium 187
7.3.2 Epsilon–Nashequilibrium 193
7.3.3 Satisfactionequilibriumandefficientsatisfactionequilibrium 195
7.3.4 GeneralizedNashequilibrium 197
7.3.5 Coarsecorrelatedequilibriumandcorrelatedequilibrium 200
7.3.6 Robustequilibrium 202
7.3.7 Bayesianequilibriumandaugmentedequilibrium 204
7.3.8 Evolutionarystablesolutions 206
Contents ix
7.3.9 Paretooptimalactionprofilesandsocialoptimalactionprofiles 210
7.3.10 Otherequilibriumconcepts 210
7.4 Learningequilibria 211
7.4.1 LearningNashequilibria 211
7.4.2 Learningepsilon-equilibrium 215
7.4.3 Learningcoarsecorrelatedequilibrium 217
7.4.4 Learningsatisfactionequilibrium 218
7.4.5 Discussion 220
7.5 Conclusion 222
References 223
PartIICognitiveradioandsharingofunlicensedspectrum 228
8 Cooperationincognitiveradionetworks:fromaccesstomonitoring 230
WalidSaadandH.VincentPoor
8.1 Introduction 230
8.1.1 Cooperationincognitiveradio:mutualbenefitsandcosts 230
8.2 Anoverviewofcoalitionalgametheory 232
8.3 Cooperativespectrumexplorationandexploitation 234
8.3.1 Motivation 234
8.3.2 Basicproblem 235
8.3.3 Jointsensingandaccessasacooperativegame 239
8.3.4 Coalitionformationalgorithmforjointsensingandaccess 241
8.3.5 Numericalresults 243
8.4 Cooperativeprimaryuseractivitymonitoring 245
8.4.1 Motivation 245
8.4.2 Primaryuseractivitymonitoring:basicmodel 246
8.4.3 Cooperativeprimaryusermonitoring 248
8.4.4 Numericalresults 255
8.5 Summary 258
Acknowledgements 259
Copyrightnotice 260
References 260
9 Cooperativecognitiveradioswithdiffusionnetworks 263
RenatoLuisGarridoCavalcante,SlawomirStan´czak,andIsaoYamada
9.1 Introduction 263
9.2 Preliminaries 264
9.2.1 Basictoolsinconvexandmatrixanalysis 265
9.2.2 Graphs 266
9.3 Distributedspectrumsensing 266
x Contents
9.4 Iterativeconsensus-basedapproaches 269
9.4.1 Averageconsensusalgorithms 269
9.4.2 Accelerationtechniquesforiterativeconsensusalgorithms 272
9.4.3 Empiricalevaluation 277
9.5 ConsensustechniquesbasedonCoMAC 280
9.6 Adaptivedistributedspectrumsensingbasedonadaptivesubgradient 284
techniques
9.6.1 Distributeddetectionwithadaptivefilters 285
9.6.2 Set-theoreticadaptivefiltersfordistributeddetection 286
9.6.3 Empiricalevaluation 293
9.7 Channelprobing 295
9.7.1 Introduction 295
9.7.2 Admissibilityproblem 296
9.7.3 Powerandadmissioncontrolalgorithms 297
9.7.4 Channelprobingforadmissioncontrol 297
9.7.5 Conclusions 299
Acknowledgements 299
References 300
10 Capacityscalinglimitsofcognitivemultipleaccessnetworks 305
EhsanNekouei,HazerInaltekin,andSubhrakantiDey
10.1 Introduction 305
10.2 Organizationandnotation 306
10.3 Threemaincognitiveradioparadigms 307
10.4 Powerallocationincognitiveradionetworks 308
10.4.1 Point-to-pointtime-invariantcognitiveradiochannels 309
10.4.2 Point-to-pointtime-varyingcognitiveradiochannels 310
10.4.3 Fadingmultipleaccesscognitiveradiochannels 311
10.5 CapacityscalingwithfullCSI:homogeneousCoEs 313
10.6 CapacityscalingwithfullCSI:heterogeneousCoEs 317
10.7 Capacityscalingwithgeneralizedfadingdistributions 318
10.8 CapacityscalingwithreducedCSI 321
10.9 Capacityscalingindistributedcognitivemultipleaccessnetworks 324
10.10Summaryandconclusions 329
Acknowledgements 331
References 332
11 Dynamicresourceallocationincognitiveradiorelaynetworksusing 334
sequentialauctions
TianyuWang,LingyangSong,andZhuHan
11.1 Introduction 334