• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Dynamic spectrum allocation and sharing in cognitive cooperative networks

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Beibei Wang
Abstract:
The dramatic increase of service quality and channel capacity in wireless networks is severely limited by the scarcity of energy and bandwidth, which are the two fundamental resources for communications. New communications and networking paradigms such as cooperative communication and cognitive radio networks emerged in recent years that can intelligently and efficiently utilize these scarce resources. With the development of these new techniques, how to design efficient spectrum allocation and sharing schemes becomes very important, due to the challenges brought by the new techniques. In this dissertation we have investigated several critical issues in spectrum allocation and sharing and address these challenges. Due to limited network resources in a multiuser radio environment, a particular user may try to exploit the resources for self-enrichment, which in turn may prompt other users to behave the same way. In addition, cognitive users are able to make intelligent decisions on spectrum usage and communication parameters based on the sensed spectrum dynamics and other users' decisions. Thus, it is important to analyze the intelligent behavior and complicated interactions of cognitive users via game-theoretic approaches. Moreover, the radio environment is highly dynamic, subject to shadowing/fading, user mobility in space/frequency domains, traffic variations, and etc. Such dynamics brings a lot of overhead when users try to optimize system performance through information exchange in real-time. Hence, statistical modeling of spectrum variations becomes essential in order to achieve near-optimal solutions on average. In this dissertation, we first study a stochastic modeling approach for dynamic spectrum access. Since the radio spectrum environment is highly dynamic, we model the traffic variations in dynamic spectrum access using continuous-time Markov chains that characterizes future traffic patterns, and optimize access probabilities to reduce performance degradation due to co-channel interference. Second, we propose an evolutionary game framework for cooperative spectrum sensing with selfish users, and develop the optimal collaboration strategy that has better performance than fully cooperating strategy. Further, we study user cooperation enforcement for cooperative networks with selfish users. We model the optimal relay selection and power control problem as a Stackelberg game, and consider the joint benefits of source nodes as buyers and relay nodes as sellers. The proposed scheme achieves the same performance compared to traditional centralized optimization while reducing the signaling overhead. Finally, we investigate possible attacks on cooperative spectrum sensing under the evolutionary sensing game framework, and analyze their damage both theoretically and by simulations.

T ABLE OF CONTENTS List of Tables vii List of Figures viii 1 Introduction 1 1.1 Motivation.................................1 1.2 Contributions and Thesis Organization.................5 2 Background 8 2.1 Related Works...............................8 2.1.1 Spectrum Sharing and Management in Cognitive Radios...8 2.1.2 Cooperative Spectrum Sensing for Primary Detection.....10 2.1.3 Relay Selection and Power Control in Cooperative Networks.11 2.2 Game-Theoretic Models.........................14 2.2.1 Stackelberg Game.........................16 2.2.2 Evolutionary Game........................17 2.3 Markov Chain...............................18 3 Primary-Prioritized Markov Approach for Dynamic Spectrum Al- location 20 3.1 System Model...............................23 3.2 Primary-Prioritized Markov Models...................27 3.2.1 CTMC without Queuing.....................27 3.2.2 CTMC with Queuing.......................33 3.3 Proposed Dynamic Spectrum Access..................36 3.4 Simulation Studies............................45 3.4.1 CTMC-8 for the Symmetric-Interference Case.........46 3.4.2 CTMC-8 for the Asymmetric-Interference Case........50 3.4.3 Comparison with a CSMA-based Scheme............52 3.4.4 Comparison with a Uniform-Access-Probability Scheme....53 3.4.5 Spectrum Sharing Among Multiple Secondary Users......54 3.5 Summary.................................55 4 Evolutionary Game for Cooperative Spectrum Sensing 57 4.1 System Model and Spectrum Sensing Game..............60 iv

4.1.1 Hypothesis of Channel Sensing.................60 4.1.2 Throughput of a Secondary User................62 4.1.3 Spectrum Sensing Game.....................63 4.2 Evolutionary Sensing Game and Strategy Analysis..................................67 4.2.1 Evolutionarily Stable Strategy..................68 4.2.2 Evolution Dynamics of the Sensing Game...........69 4.2.3 Sensing Game with Homogeneous Players...........71 4.2.4 Sensing Game with Heterogeneous Players...........76 4.2.5 Learning Algorithm for ESS...................80 4.3 Simulation Studies............................83 4.3.1 Sensing Game with Homogeneous Players...........83 4.3.2 Convergence of the Dynamics..................85 4.3.3 Comparison of ESS and Full Cooperation............87 4.4 Summary.................................88 5 Stackelberg Game for Distributed Resource Allocation in Cooper- ative Networks 90 5.1 System Description............................92 5.1.1 System Model...........................92 5.1.2 Problem Formulation.......................96 5.2 Analysis of the Proposed Games.....................99 5.2.1 Buyer-Level Game for the Source Node.............99 5.2.2 Seller-Level Game for the Relay Nodes.............102 5.2.3 Existence of the Equilibrium...................103 5.2.4 Distributed Price Updating...................109 5.2.5 Comparison with the Centralized Optimal Scheme.......115 5.3 Simulation Studies............................119 5.3.1 One-Relay Case..........................120 5.3.2 Two-Relay Case..........................122 5.3.3 Multiple-Relay Case.......................124 5.3.4 Convergence Speed of the Game.................126 5.3.5 Comparison with the Centralized Optimal Scheme.......128 5.3.6 Effect of the Bandwidth Factor.................130 5.4 Summary.................................132 6 Attacks in Spectrum Sensing 133 6.1 Mask Primary User Signal........................133 6.2 Report Faulty Sensory Data.......................135 6.3 Simulation Studies............................138 6.4 Summary.................................143 7 Conclusions and Future Work 144 7.1 Conclusions................................144 7.2 Future Work................................147 v

Bibliograph y 149 vi

LIST OF TABLES 3.1 Primary-prioritized dynamic spectrum access..............42 4.1 Payoff table of a two-user sensing game.................77 vii

LIST OF FIGURES 3.1 System model (upper:system diagram;lower:throughput vs.time).24 3.2 The rate diagram of CTMC with no queuing..............28 3.3 The rate diagram of CTMC with queuing................33 3.4 Modified CTMC with access control (no queuing)............38 3.5 Access probability vs.λ A (symmetric-interference,λ B = 85 s −1 )....47 3.6 Average throughput vs.λ A (symmetric-interference,λ B = 85 s −1 )...48 3.7 Access probability for different λ P (λ B = 85 s −1 )............49 3.8 Access probability vs.λ A (asymmetric-interference,λ B = 85 s −1 )...50 3.9 Average throughput vs.λ A (asymmetric-interference,λ B = 85 s −1 )..51 3.10 Overall throughput for CTMC-5,CTMC-8 and CSMA.........52 3.11 The histogram of throughput improvement (U PF )...........54 3.12 Comparison of overall throughput for multiple secondary users.....55 4.1 System model...............................64 4.2 Cooperative spectrum sensing......................65 4.3 ESS and average throughput vs.τ....................84 4.4 Behavior dynamics of a homogeneous K-user sensing game......86 4.5 Behavior dynamics of a heterogeneous 3-user sensing game......87 4.6 Comparison of ESS and full cooperation................88 viii

5.1 System diagrams..............................93 5.2 Comparison of optimal relay power of the game and the centralized scheme...................................118 5.3 1-relay case with the relay node at different locations..........121 5.4 2-relay case with relay node r 2 at different locations..........123 5.5 Multiple-relay case with different number of relay nodes........125 5.6 Observation of convergence speed.....................127 5.7 Optimal rate in distributed and centralized schemes..........129 5.8 Optimal U s including the bandwidth-factor effect,with different relay nodes’ help,a = 0.85............................131 6.1 ESS and average throughput vs.τ when the primary signal is masked.139 6.2 ESS and average throughput vs.τ when 2 malicious users report faulty sensory data.............................140 6.3 ESS and average throughput vs.τ when 3 malicious users report faulty sensory data.............................142 ix

Chapter 1 Introduction 1.1 Motivation The dramatic increase of service quality and channel capacity in wireless net- works is severely limited by the scarcity of energy and bandwidth,which are the two fundamental resources for communications.Therefore,researchers are currently focusing their attention on new communications and networking paradigms that can intelligently and efficiently utilize these scarce resources.For instance,cooperative communications [LTW04] can take advantage of the broadcasting nature of wire- less networks and exploit the inherent spatial and multiuser diversities,where relay nodes act as a virtual antenna array to help source nodes forward information to the destination nodes to achieve higher data throughput and more reliable transmis- sion.Moreover,with the development of cognitive radio technology [III00],future wireless communication devices are envisioned to be able to sense and analyze their surrounding environment,learn from the environment variations,and adapt their 1

op erating parameters accordingly in order for a better performance and more ef- ficient spectrum utilization.For instance,when cognitive network users share a licensed spectrum,they can detect spectrum white space,select the best frequency bands,coordinate spectrum access with other users and vacate the frequency when a primary user appears. In traditional military and emergency applications,users in a wireless net- work usually belong to the same authority and have the same objective.However, in emerging networks,such as cooperative and cognitive communication networks envisioned in civilian applications,different network users typically belong to dif- ferent operators and may pursue different goals.Fully cooperative behaviors such as forwarding data for other users or contributing to a common task uncondition- ally cannot be pre-assumed.Due to limited network resources in a multiuser radio environment,a particular user may try to exploit the resources for self-enrichment, which in turn may prompt other users to behave the same way.Moreover,the ra- dio environment is highly dynamic,subject to shadowing/fading,user mobility in space/frequency domains,traffic variations,and etc.Such dynamics brings a lot of overhead when users try to optimize system performance through information exchange in real-time. Since these emerging communication paradigms are usually deployed in a highly dynamic spectrumenvironment where network users tend to be selfish,before they can be successfully exploited in order to achieve efficient spectrum utilization, the following two critical issues must be resolved first:user cooperation and dy- namic spectrum sharing.Since selfish network users only aim at maximizing their 2

o wn benefits,without a properly designed cooperation enforcement mechanism,they may be unwilling to help other users at the expense of their own resources,such as forward data for other users or spend their own time detecting spectrum white space.Furthermore,dynamics caused by different user activities,e.g.,primary users re-occupying/vacating their licensed bands and secondary users starting/ceasing a communication session,pose even greater challenges for the design of dynamic spec- trum sharing schemes. In recent years,efficient spectrum allocation and sharing in cooperative and cognitive communication networks has drawn extensive attentions [CZ05,ZC05, EPT07,JL07,JL06,RYM05,SA06].The performance in cooperative communica- tion networks depends on careful resource allocations such as relay placement,re- lay selection,bandwidth allocation,and transmission power control.Transmission power allocation is optimized to minimize the outage probability and maximize network lifetime [HA03,SSL08].Relay selection and assignment schemes are pro- posed to fully utilize the cooperative diversity,minimize the outage probability, extend coverage area,and maximize throughput [LBG + 04,BLR05,SHL06].In ad- dition,there have been several previous efforts addressing how to efficiently and fairly share the spectrum resources in cognitive communication networks,on a ne- gotiation/pricing basis [CZ05,ZC05,EPT07,JL07,JL06,RYM05] or an opportunistic basis [XCMS06,KC06].In negotiation/pricing-based spectrum allocation,the un- used spectrumresources fromlegacy spectrumholders (primary users) can be shared among unlicensed users through auction-based pricing approaches.In opportunis- tic spectrum sharing,unlicensed/secondary users can access the licensed spectrum 3

when the spectrum is sensed as idle.Furthermore,in order to protect primary users from the interference due to secondary spectrum usage,various spectrum detection schemes have been proposed to improve the detection performance and minimize the conflict/interference with the primary user.Recent study has shown that co- operative spectrum sensing with multiple secondary users can further improve the efficiency of primary user detection. Although the existing spectrum allocation and sharing schemes can enhance system performance in cooperative and cognitive communication networks,there are still some fundamental issues that require further treatment. First,the radio spectrum environment is constantly changing.In conventional power control to manage mutual interference for a fixed number of secondary users, after each change of the number of contending secondary users,the network needs to re-optimize the power allocation for all users completely.This results in high complexity and much overhead.If a primary user appears in some specific portion of the spectrum,secondary users in that band also need to adapt their transmission parameters to avoid interfering with the primary user.Therefore,efficient dynamic spectrum sharing scheme must include a traffic model based on traffic statistics to predict the future traffic patterns in the shared spectrum. Second,in order to improve the detection performance of a primary user, most cooperative spectrum sensing schemes assume a fully cooperative scenario, meaning all secondary user will voluntarily fuse their sensing outcomes to a common controller such as a secondary base station.But this assumption does not hold in a decentralized network.Moreover,due to users’ specific channel conditions,it is even 4

not optimal to have all secondary users cooperate in every sensing effort.In addition, sensing takes energy/time which may be diverted to useful data transmission.In self-organizing networks where secondary users exchange sensory data to make a final decision,selfish users tend to take advantage of the others so as to reserve more time for their own data transmission.Therefore,how to collaborate with selfish users in cooperative spectrum sensing is another very important issue. Third,most existing works on spectrumsharing in cooperative communication networks mainly focus on resource allocation by means of a centralized fashion. Such schemes require that complete and precise channel state information (CSI) be available in order to optimize the system performance,which are generally neither scalable nor robust to channel estimation errors.Moreover,users in decentralized self-organizing cooperative communication networks belong to different authorities. Therefore,a mechanism of reimbursement to relay nodes is needed such that relay nodes can earn benefits from spending their own transmission power in helping the source node forward information. 1.2 Contributions and Thesis Organization This dissertation has investigated how to efficiently utilize the limited network resources in cognitive cooperative networks with selfish users under a time-varying spectrumradio environment.Specifically,two important issues have been addressed: 1) how to collaborate with selfish users in forwarding information and cooperative spectrum sensing,and 2) how to design dynamic spectrum access strategy by uti- 5

lizing the traffic statistics to predict future traffic patterns and reduce information exchange.The contributions lie in the following three aspects. Evolution of behavior dynamics in cooperative spectrum sensing:In order to study the time evolution of selfish users’ cooperation behavior,we have proposed an evolutionary cooperative sensing game,derived users’ behavior dynam- ics,and proved their convergence to the evolutionarily stable strategy (ESS).The proposed approach not only reveals the underlying behavior dynamics involved in establishing robust equilibrium,but also helps to develop a distributed learning algorithm that guides secondary users to approach the ESS only with their own throughput observation.More important,it opens a new avenue for future research on studying behavior dynamics in cognitive radio networks using evolutionary game theory. Statistical modelling for dynamic spectrum access:Another contribu- tion of this dissertation is the traffic modelling of primary and secondary users in cognitive radio networks.Specifically,we have modelled the traffic variations of the radio environment as continuous-time Markov chains (CTMC).Since the model can characterize the traffic dynamics of different users occupying the licensed spectrum, the proposed approach provides a means for predicting future traffic patterns in the shared spectrum.As mutual interference will impair spectrum efficiency when multiple secondary users transmit in the same frequency,in order to compensate throughput degradation due to mutual interference,we have further introduced op- timal access probabilities for secondary users so that the chance of spectrumsharing is controlled and spectrum resources are shared in a more efficient way without con- 6

flicting with primary users. User cooperation enforcement in cooperative networks:In this dis- sertation,we have also proposed a two-level Stackelberg game which considers the joint benefits of source nodes as buyers and relay nodes as sellers.With the pro- posed approach,not only can source nodes and relays at relatively better locations and buy an optimal amount of power,but also competing relays can maximize their profits by asking optimal prices.Furthermore,we have designed a distributed price updating function by which relay nodes can iteratively approach their optimal prices and system performance is gradually optimized.Compared to most existing works, the proposed approach does not require CSI,and therefore greatly reduces overhead and signaling.Moreover,the distributed nature of the proposed scheme makes it a building block in large-scale wireless ad hoc networks for the cooperation simulation among nodes. The reminder of this dissertation is organized as follows.Chapter 2 introduces the related works,game-theoretic models,and basic concepts of Markov chain for spectrum allocation and sharing in cognitive cooperative networks.In Chapter 3, a primary-prioritized Markov approach is described for dynamic spectrum access in licensed bands [WJLC09].In Chapter 4,an evolutionary game framework for cooperative spectrum sensing with selfish users is discussed [WLC09].The user collaboration enforcement in cooperative wireless networks using Stackelberg game [WHL09] is presented in Chapter 5.Malicious attacks on cooperative spectrum sensing are studied in Chapter 6.Finally,Chapter 7 concludes this dissertation and discusses the future work. 7

Chapter 2 Background 2.1 Related Works 2.1.1 Spectrum Sharing and Management in Cognitive Ra- dios The usage of radio spectrumresources and the regulation of radio emissions are coordinated by national regulatory bodies like the Federal Communications Com- mission (FCC).The FCC assigns spectrum to license holders or services on a long- term basis for large geographical regions;however,a large portion of the assigned spectrumremains unutilized.The inefficient usage of the limited spectrumresources necessitates the development of dynamic spectrum access techniques.Recently,the FCC began considering more flexible and comprehensive uses of the available spec- trum [FCC02,FCC03b],through the use of cognitive radio technology [III00].By exploiting the spectrum in an opportunistic fashion,dynamic spectrum access en- 8

ables secondary users to sense which portions of the spectrum are available,select the best channel,coordinate access to spectrum channels with other users,and vacate the channel when a primary user appears. In order to fully utilize the limited spectrum resources,how to efficiently and fairly share the spectrum among secondary users becomes an important issue,es- pecially when multiple dissimilar secondary users coexist in the same portion of the spectrum band.There have been several previous efforts addressing this issue on a negotiated/pricing basis [CZ05,ZC05,EPT07,JL07,JL06,RYM05] or an opportunis- tic basis [XCMS06,KC06].A local bargaining mechanismwas proposed in [CZ05] to distributively optimize the efficiency of spectrumallocation and maintain bargaining fairness among secondary users.In [HBH06],auction mechanisms were proposed for sharing spectrumamong multiple users such that the interference was controlled be- low a certain level.Rule-based approaches were proposed in [ZC05] that regulated users’ spectrum access in order to trade-off fairness and utilization with commu- nication costs and algorithmic complexity.In [EPT07],the authors proposed a repeated game approach,in which the spectrum sharing strategy could be enforced using the Nash Equilibrium of dynamic games.In [JL07,JL06],belief-assisted dy- namic pricing was used to optimize the overall spectrum efficiency while basing the participating incentives of the selfish users on double auction rules.A centralized spectrum server was considered in [RYM05] to coordinate the transmissions of a group of wireless links sharing a common spectrum.Recently,attention is being drawn to opportunistic spectrum sharing.In [XCMS06],a distributed random ac- cess protocol was proposed to achieve airtime fairness between dissimilar secondary 9

users in open spectrum wireless networks without considering primary users’ activ- ities.The work in [KC06] examined the impact of secondary user access patterns on blocking probability and achievable improvement in spectrum utilization with statistical multiplexing,and proposed a feasible spectrum sharing scheme. 2.1.2 Cooperative SpectrumSensing for Primary Detection In order to identify the spectrum hole when the primary is inactive,an impor- tant requirement of secondary users is the capability to sense their surrounding radio spectrum environment.Further,since primary users should be carefully protected from interference due to secondary users’ operation,secondary users also need to sense the licensed spectrumbefore each transmission and can only transmit when the spectrum is idle.One efficient way of spectrum detection is the primary transmitter detection based on local observations of secondary users.If the information of the primary user signal is known to secondary users,they can use matched filter detec- tion to maximize the received signal-to-noise ratio (SNR) under stationary Gaussian noise [SC05].Another detection approach is the cyclostationary feature detection which analyzes the spectral correlation function of modulated signals [FGR05].Al- though these two types of detection schemes can achieve precise detection,they either require priori knowledge of the primary user signal,or have high complexity. A good alternative is energy detection. Energy detection is suitable to the scenario when secondary users cannot gather sufficient information about the primary user signal.An energy detector 10

collects locally observed signal samples within a certain time,measures the signal energy,and compare the energy with a threshold to determine whether a primary user is present or not [SC05].However,energy detection is heavily affected by noise uncertainty. Recently,cooperative spectrum sensing with relay nodes’ help and multi- user collaborative sensing has been shown to greatly improve the sensing perfor- mance [GS05,MSB06,GL07,PL07,LYZH07,VJP08,GLBL08].In [GS05],the au- thors proposed collaborative spectrum sensing to combat shadowing/fading effects. The work in [MSB06] proposed light-weight cooperation in sensing based on hard decisions to reduce the sensitivity requirements.The authors of [GL07] showed that cooperative sensing can reduce the detection time of the primary user and increase the overall agility.How to choose proper secondary users for cooperation was in- vestigated in [PL07].The authors of [LYZH07] studied the design of sensing slot duration to maximize secondary users’ throughput under certain constraints.Two energy-based cooperative detection methods using weighted combining were ana- lyzed in [VJP08].Spatial diversity in multiuser networks to improve spectrum sens- ing capabilities of centralized cognitive radio networks were exploited in [GLBL08]. 2.1.3 Relay Selection and Power Control in Cooperative Networks The performance in cooperative communication networks depends on care- ful resource allocation,such as relay placement,relay selection,and power con- 11

trol [HA03,SSL08,MY04,LBG + 04,BLR05,SHL06,HHSL05,ISSL08,ZAL06,NY07, SS07,HSHL07,ACM07,LHXS07].In [HA03],the power allocation was optimized to satisfy the outage probability criterion.The authors in [SSL08] provided the analysis on symbol error rates and optimum power allocations for the decode-and- forward cooperation protocol in wireless networks.The energy-efficient broadcast problem in wireless networks was considered in [MY04].The work in [LBG + 04] evaluated the cooperative diversity performance when the best relay was chosen according to the average SNR,and the outage probability of relay selection based on instantaneous SNRs.In [BLR05],the authors proposed a distributed relay se- lection scheme that required limited network knowledge with instantaneous SNRs. In [SHL06],the relay assignment problem was solved for the multiuser cooperative communications.In [HHSL05],the cooperative resource allocation for OFDM was studied.The authors of [ISSL08] investigated the relay selection problem with focus on when to cooperate and which relay to cooperate with,which required channel state information (CSI).In [ZAL06],centralized power allocation schemes were pre- sented by assuming all the relay nodes should help.In order to further minimize the system outage behaviors and improve the average throughput,a selection forward protocol was proposed to choose only one “best” relay node to assist transmission. A centralized resource allocation algorithm for power control,bandwidth allocation, relay selection and relay strategy choice in an OFDMA-based relay network was proposed in [NY07].The work in [SS07] developed distributed power control strate- gies for multi-hop cooperative transmission schemes.Lifetime extension for wireless sensor networks with the aid of relay selection and power management schemes was 12

in vestigated in [HSHL07].The authors of [ACM07] studied the optimal power al- location problem in the high SNR regime for different relaying protocols.Relay station placement and relay time allocation in IEEE 802.16j networks was investi- gated in [LHXS07]. In recent years,some efforts have been made towards mathematical analysis of cooperation in wireless networking using game theory,since game theory is a natural and flexible tool that studies how the autonomous network users interact and cooperate with each other.In game-theoretic literature of wireless networking, in [MW01],the behaviors of selfish nodes in the case of random access and power control were examined.In [DPA00],static pricing policies for multiple-service net- works were proposed.Such policies can offer incentives for each node to choose the service that best matches its needs,so as to discourage over-allocation of resources and improve social welfare.The work in [SMG02] presented a power control solution for wireless data in the analytical setting of a game-theoretic framework.Pricing of transmit powers was introduced to improve user utilities that reflected the quality of service a wireless terminal received.A pricing game that stimulated cooperation via reimbursements to the relay was proposed in [SA06],but there was no detailed analysis on how to select the best relays and how to achieve the equilibrium dis- tributively.In [HJL05],the authors employed a cooperative game for the single-cell OFDMA resource allocation. 13

2.2 Game-Theoretic Models In self-organized wireless networks,users may belong to different operators and compete for limited network resources,In other words,users are selfish and only aim at maximizing their own profits by utilizing more resources.With the rapid development of cognitive radio technique,network users can further make in- telligent decisions on spectrum usage and communication parameters based on the sensed spectrum dynamics and other users’ decisions.Moreover,many fundamental network functionalities,such as packet forwarding,relaying information,and spec- trumsharing,cannot be performed without relying on cooperation among the selfish users.In such scenarios,it is no longer feasible to optimize the network performance by assuming there is a central authority and every user will obey the resource alloca- tion rule.Therefore,it is natural to study the intelligent behaviors and interactions of selfish network users from the game-theoretic perspective. Game theory [FL91,FT93] is a mathematical tool that analyzes the strategic interactions among rational decision makers.Three major components in a strategic- form game model are the set of players,the strategy space of each player,and the payoff/utility function,which measures the outcome of the game for each player. In cooperative communication networks,source nodes may need to provide relay nodes with incentives for relaying their data and relay nodes need to select the best pricing strategy,which can be modeled as a buyer/seller game [WHL07, WHL09,SA06,HL08].In cognitive radio networks,the competition and cooperation among the cognitive network users can also be well modeled as a spectrum sharing 14

game [WWLC09,EPT07,JL07,HBH06,Y.09,WWJ + 08].Specifically,in open spec- trum sharing (e.g.,[WWLC09,EPT07]),the players are all the secondary users that compete for unlicensed spectrum;in licensed spectrumsharing,where primary users lease their unused bands to secondary users,the players include both the primary and secondary users (e.g.,[HBH06,Y.09,JL06]). The strategy space for each player may vary according to the specific spectrum sharing scenario.For instance,in cooperative communication networks,the strategy space of the source nodes includes which relay to choose for help and the relay power levels,while the strategy space of the relay nodes are the prices for relaying data. In open spectrum sharing,the strategy space of secondary users may include the transmission parameters they want to adopt,such as the transmission powers,access rates,time duration,etc.;while in licensed spectrum trading,their strategy space includes which licensed bands they want to rent,and how much they would pay for leasing those licensed bands.For the primary users,the strategy space may include which secondary users they would lease each of their unused bands to,and how much they will charge for each band. The payoff functions for different users are defined to characterize various performance criteria accordingly.For instance,in open spectrumsharing,the payoff function for the secondary users is often defined as a non-decreasing function of the Quality of Service (QoS) they receive by utilizing the unlicensed band;in licensed spectrum trading,the payoff function for the users often represents the monetary gains (e.g.,revenue minus cost) by leasing the licensed bands. In a noncooperative spectrum sharing game with selfish network users,each 15

user only targets for maximizing his/her own payoff by choosing an optimal strat- egy.And the outcome of the noncooperative game is often measured by the Nash Equilibrium (NE).The NE is defined as the set of strategies for all the users such that no user can improve his/her payoff by unilaterally deviating from the equilib- rium strategy,given that the other users adopt the equilibrium strategies.So the NE indicates that no individual user would have the incentive to choose a different strategy. 2.2.1 Stackelberg Game If players choose their strategies simultaneously,the game can be described using a strategic-form representation [FT93].In order to model a game with a dynamic structure,game theorists use the concept of a game in extensive form, which explicitly states the order in which players move,and what each player knows when making each of his decisions [FT93].An example of an extensive-form game is the Stackelberg game [FT93].In a Stackelberg game,one player must commit to a strategy before other players choose their own strategies.Specifically,the players of a Stackelberg game include a leader and a follower/followers.A leader commits to a strategy first,and then a follower selfishly optimizes his own reward, considering the strategy selected by the leader.Compared to Nash games,where all players take their moves simultaneously,Stackelberg games can better model the scenario with heterogeneous players that take sequential moves.For example in cooperative communication networks with self-organized nodes/users,after a source 16

no de broadcasts his data to a destination node and several relay nodes,the relay nodes will signal their optimal prices for relaying data to the source node,and the source node then optimizes the service he purchases from the relay nodes.Thus, we can use Stackelberg game to model such cooperation,where the source node is the leader and the relay nodes are the followers.Stackelberg game have been used to model congestion control [BS02],attacker-defender scenarios in security domains [PPM + 08],network routing [KLO97] and scheduling strategies [Rou01]. 2.2.2 Evolutionary Game There can be more than one NE in a game.When there exist several different NE,howshould a rational player decide which of themis the “right one”?Game the- orists have proposed different refinement criteria [FT93];however,each equilibrium could be justified by some refinement.The problembecomes even more complicated if the players are uncertain about the game being played and the game involves a dynamic process.Therefore,evolutionary game theory (EGT) was proposed [Smi82] to reveal the underlying dynamics and find a robust equilibrium. The idea of EGT was inspired by the study of ecological biology,and it differs from classic game theory by focusing on the dynamics of strategy change more than the properties of strategy equilibrium.EGT was first used to study the adjustment of population fractions by evolution,which states that the genes whose strategies are more successful will have higher reproductive fitness.Therefore,the population fractions of strategies whose payoff against the current distribution of opponents’ 17

Full document contains 173 pages
Abstract: The dramatic increase of service quality and channel capacity in wireless networks is severely limited by the scarcity of energy and bandwidth, which are the two fundamental resources for communications. New communications and networking paradigms such as cooperative communication and cognitive radio networks emerged in recent years that can intelligently and efficiently utilize these scarce resources. With the development of these new techniques, how to design efficient spectrum allocation and sharing schemes becomes very important, due to the challenges brought by the new techniques. In this dissertation we have investigated several critical issues in spectrum allocation and sharing and address these challenges. Due to limited network resources in a multiuser radio environment, a particular user may try to exploit the resources for self-enrichment, which in turn may prompt other users to behave the same way. In addition, cognitive users are able to make intelligent decisions on spectrum usage and communication parameters based on the sensed spectrum dynamics and other users' decisions. Thus, it is important to analyze the intelligent behavior and complicated interactions of cognitive users via game-theoretic approaches. Moreover, the radio environment is highly dynamic, subject to shadowing/fading, user mobility in space/frequency domains, traffic variations, and etc. Such dynamics brings a lot of overhead when users try to optimize system performance through information exchange in real-time. Hence, statistical modeling of spectrum variations becomes essential in order to achieve near-optimal solutions on average. In this dissertation, we first study a stochastic modeling approach for dynamic spectrum access. Since the radio spectrum environment is highly dynamic, we model the traffic variations in dynamic spectrum access using continuous-time Markov chains that characterizes future traffic patterns, and optimize access probabilities to reduce performance degradation due to co-channel interference. Second, we propose an evolutionary game framework for cooperative spectrum sensing with selfish users, and develop the optimal collaboration strategy that has better performance than fully cooperating strategy. Further, we study user cooperation enforcement for cooperative networks with selfish users. We model the optimal relay selection and power control problem as a Stackelberg game, and consider the joint benefits of source nodes as buyers and relay nodes as sellers. The proposed scheme achieves the same performance compared to traditional centralized optimization while reducing the signaling overhead. Finally, we investigate possible attacks on cooperative spectrum sensing under the evolutionary sensing game framework, and analyze their damage both theoretically and by simulations.