Intelligent clustering cooperative spectrum sensing based on Bayesian learning for cognitive radio network

Abstract

In cognitive radio network, cooperative spectrum sensing (CSS) can improve the sensing performance on the absence of a primary user, when the sensing channel is in severe fading and shadowing effect. However, CSS is sensitive to the fading reporting channel. In this paper, an intelligent clustering CSS based on Bayesian learning is proposed to improve sensing performance under both perfect and imperfect sensing reports as well as decrease the rate loss and cooperative overhead. The clustering CSS is performed by intra-cluster CSS and inter-cluster CSS. An optimal sensing threshold for the intra-cluster CSS is achieved by minimizing the total Bayesian cost. The total false alarm probability and detection probability for the inter-cluster CSS are obtained by the Bayesian fusion. A clustering algorithm based on K-means learning is proposed to classify the sensing nodes and select the cluster heads. The simulation results have shown that the proposed clustering CSS outperforms the traditional CSS without clustering in the aspects of sensing performance and time overhead.