An adaptive channel selection method based on available channel duration prediction for wireless cognitive networks

Cognitive radio makes ad-hoc network take full use of underutilized spectrum resources and improve performance of transmission. However, because of communication traffic increasing and time-variance of channels in ad-hoc network, errors exist in spectrum prediction, which makes collision unavoidable during channel selection. Therefore, we introduce theories of relevance vector machine and propose a method to predict probability interval of channel state duration based on cognitive radio technologies. We then proposed a method to allocate and switch channels, in which we take the prediction results as decision basis. According to model building and simulation of the ad-hoc network system, it is proved that algorithm proposed can lower average collision ratio between users and interference and guarantee delay of channel switchover and throughput of the system at the same time. Besides, the algorithm can lower interference collision ratio efficiently even when the network gets saturation.