Evolutionarily stable opportunistic spectrum access in cognitive radio networks

Abstract

In order to fully utilise limited spectrum resources of multiple channels and multiple radios in cognitive radio networks, the authors propose a potential game model for opportunistic spectrum access based on both accurate and inaccurate spectrum state estimation with considering the interference constraints of licensed users. Three algorithms are proposed to achieve equilibrium of the proposed game. First, assuming spectrum sensing results are accurate, a joint strategy fictitious play-based channel selection algorithm with incomplete information is presented, and it can achieve a pure Nash equilibrium (NE) of the proposed game. Second, in order to make the outcomes of game robust, an evolutionary spectrum access mechanism with complete information is introduced by using evolutionary game theory based on inaccurate spectrum state estimation so that evolutionary stable strategy (ESS) can be achieved. Finally, with incomplete network information, a distributed learning algorithm is proposed to achieve a mixed NE, which is proved to be an ESS. Simulation results show that these algorithms can significantly improve spectrum allocation efficiency while reducing mutual collision.