OPTIMAL DESIGN FOR DYNAMIC SPECTRUM ACCESS IN COGNITIVE RADIO NETWORKS UNDER RAYLEIGH FADING

Abstract

We consider a time slotted cognitive radio network under Rayleigh fading where multiple secondary users (SUs) contend for spectrum usage over available primary users' channels. We analyze the performance of a channel access policy where each SU stochastically determines whether to access a wireless channel or not based on a given access probability. In the analysis, we focus on the queueing performance of an arbitrary SU with the channel access policy. To improve the queueing performance of SUs, the access probability in our channel access policy is adapted to the knowledge on the wireless channel information, e.g., the number of available channels and the nonfading probability of channels. It is then important to obtain the optimal access probabilities from the queueing performance perspective. In this paper we consider three scenarios. In the first scenario, all SUs have full information on wireless channel status and fading channel conditions. In the seconds cenario, all SUs have the information on wireless channel status but do not know their fading channel conditions, and in the last scenario all SUs do not have any information on wireless channel status and conditions. For each scenario we analyze the queueing performance of an arbitrary SU and show how to obtain the optimal access probabilities with the help of the effective bandwidth theory. From our analysis we provide an insight on how to design an optimal channel access policy in each scenario. We also show how the optimal channel access policies in three scenarios are related with each other. Numerical results are provided to validate our analysis. In addition, we investigate the performance behaviors of the optimal channel access policies.