Stochastic spectrum access based on learning automata in cognitive radio network

We consider dynamic spectrum access among cognitive radio from an adaptive learning perspective. In order to avoid the costly channel switching and to ensure QoS satisfaction of nodes, a secondary user may desire an optimal channel which maximizes the throughput, rather than consistently adapting channels to the random environment. We propose a stochastic spectrum access based on learning automata which takes into account the collision probability and channel quality simultaneously. The algorithm would track the variation of channels without prior knowledge of environment required and converge to the ¿-optimal solution asymptotically. This procedure is shown to perform very well compared with other similar adaptive algorithms in numerical simulations.