The Key Technology Of Spectrum Sharing In Non-cooperative Cognitive Radio Networks

Cognitive radio networks(CRN)have aroused increasing attentions of academia and industry because the spectrum efficiency can be improved largely with dynamic spectrum access.Spectrum sharing is the pivotal research content of CRN,and it focuses mainly on spectrum acquisition and exploitation among primary users(PU)and secondary users(SU).This dissertation considers this problem in nocooperative scenario and studies the interaction of rational and selfish users.We research spectrum sharing from three aspects,including spectrum acquisition,channel shaping and spectrum access.For the spectrum acquisition,we study how to make more secondary users with budget constraints obtain spectrum through spectrum trading with primary users.After obtain spectrum resource,we consider the channel shaping problem and channel assignment among secondary users.At last,we study the problem of high efficiency access to the channel established by each user.The main contributions of this dissertation include:(1)Spectrum trading model and algorithm for secondary users with budget constraints.Spectrum trading is one of the most important manners of spectrum acquisition,however,the influence of budget on the trading is barely considered in related works.Since spectrum resource is very expensive,SU may not be able to afford alone,which could lead to spectrum shortage.To resolve the problem of low transaction rate,we divide the spectrum trading market into two tier markets and design two novel spectrum trading mechanisms in each market.The main idea for these two mechanisms is that multiple users buy spectrum jointly and sharing the cost of trading.In the first tier market we design a group-buying model called COSTAG between primary user and secondary users.A new grouping rule is proposed to select the users to attend group-buying and a pricing rule is designed to decide how much each user should pay for spectrum.COSTAG can ensure the group-strategy proof,increase successful rate and the revenue of PU.In the second tier market we design the spectrum sublet game SSG among SUs so that multiple SUs can share spectrum trading cost by sublet.The game relation is built by using multi-leader multi-follower game model,and we prove the existence of Nash equilibrium(NE)through exploiting shared constraint strategy space and design distributed algorithm to approach NE.(2)Strategy research of user competition through channel allocation.The channel of SU is usually assigned by service provider(SP)in infrastructure-based networks.Different channel assignment schemes result in different revenue for SP,which lead to the competition for SUs between SPs.Pricing scheme is often used to resolve the competition among SPs in existing works,but drastic pricing war may harms SP's revenue,and furthermore,the common pricing scheme could lead to inefficiency of resource allocation.A novel user competition mechanism based on channel pre-allocation is proposed in this dissertation,and we extend Blotto game to model this competition game to study the optimal SP's strategy with the constraint of limited spectrum and heterogeneous channel gains.We give the conditions of the existence of NE,and prove the existence of mixed NE by exploiting complementary continuous of payoff function.At last,we propose an algorithm EFPA based on fictitious play to converge to mixed NE for SPs,and the iteration of the solution for EFPA is constructed through dynamic program for multidimensional nonzero game.(3)Designing the spectrum sharing game with channel reconstruction.Each SU selects its channel with different bandwidth and frequency on its own in the noncooperative Ad Hoc networks.In this case,SU can occupy more spectrum to increase data rate,which lead to ”tragedy of the commons”.Global controller or pricing scheme are usually used in related works to restrict excess channel selection,but it's not easy to implement this scheme in noncooperative networks.We study the problem of how to construct appropriate channel with different bandwidth and frequency for each SU,and an adaptive spectrum allocation model is proposed based on noncooperative game theory.The strategy of each SU is independently to maximize its payoff.Different from existing works,the payoff function of player is constructed by cross-lay design without external intervention.By reducing this game model to potential game,we prove the existence of NE,and the algorithm for approaching NE is based on improvement response.Simulation results demonstrate the effectiveness and convergence of this algorithm.(4)Channel reconstruction-oriented medium access control protocol CH-MAC.SUs construct channels with different bandwidth and frequency to communicate with access point in an infrastructure-based network.In this way,not only the spectrum efficiency can be improved,the problem of the lack of common data channel in traditional single channel access can also be resolved.However,this kind of heterogeneous multi-channel access is difficult to implement.It could be realized with strict time synchronization and centralized control in cell network,but it's not easy in asynchronous network,especially when the channel is selected by end users.So we design a new dynamic multi-channel access control protocol,called CH-MAC in cognitive radio networks,which can resolve the channel negotiation,multi-channel contention and conflict detection.We exploit subcarrier coding to send RTS to make AP with single radio receive packets on multiple channels simultaneously.The feasibility and performance of CH-MAC is tested through implementing it with cognitive radio testbed and simulation.