Distributed cooperation based Opportunistic Spectrum Access

Abstract

Opportunistic Spectrum Access (OSA), where cognitive radios opportunistically access spectrum bands currently unused by primary users, can significantly enhance the spectrum utilization. The more channels sensed, the more spectrum opportunities. As such, the more sensing time spent, the less transmission time. Thus, there is a tradeoff between spectrum sensing and data transmission. A natural question that arises then is how many and which channels should be sensed and accessed, respectively, to maximize system throughput. Considering sensing overhead, this paper develops an analytical framework for distributed cooperation based OSA by using Partially Observable Markov Decision Process (POMDP). A suboptimal greedy strategy with reduced complexity yet comparable performance is also developed.