Distributed stochastic learning for dynamic spectrum access adaptive to primary network conditions

We consider the problem of decentralized online learning and channel access among M secondary users (SUs) in a cognitive radio network. We aim at designing an adaptive policy that can effectively respond to different primary network conditions. By applying stochastic learning automata, we propose an adaptive decentralized access policy. Each SU probabilistically chooses one of the M-best channels to access. The channel selection probability is then updated based on collision events. Our proposed adaptive policy utilizes two underlying distributed learning algorithms: one is to learn from sensing history on the primary channel availability, and the other is to learn from collision history on channel selections among SUs to avoid further collision. Some previously proposed distributed access policies can be viewed as special cases of our proposed adaptive policy, with a set of pre-set channel selection probabilities. Simulation results demonstrate the effectiveness of our proposed adaptive policy in various distributions of mean channel availabilities across primary channels, as compared with other existing policies.