Dynamic Opportunistic Spectrum Access of Multi-channel Multi-radio Based on Game Theory in Wireless Cognitive Network

Abstract

Using partial observable Markov decision process(POMDP) and game theoretic solutions, we investigate the problem of achieving global optimization for distributed channel selections in cognitive radio networks (CRNs). In order to fully utilize the scarce spectrum resources, we propose two special cases to study the dynamic spectrum access. Firstly, the channel state prediction based on POMDP could reduce the collision of SUs with PUs, Secondly, a potential game(PG) theoretic framework and joint strategy fictitious play(JSFP) have been proposed to determine the access probability of SU, which could reduce the collision with other SUs. It is shown that with the proposed cases, global optimization has been achieved with local information. Specifically, the strategy with two cases mentioned above maximizes the network throughput and minimizes the network collision level. Meanwhile, the JSFP, which works to provides strong guarantees on the resulting asymptotic behavior, is proposed to achieve the global optimum autonomously and rapidly. Simulation results show that the proposed scheme can greatly improve the spectrum efficiency by alleviating mutual collision.