Centralized and decentralized cooperative spectrum sensing in cognitive radio networks: A novel approach

Abstract

In this paper, the cooperative spectrum sensing is probabilistically modeled as a mixture of two Gaussian distributions and EM algorithm is applied for learning the parameters and classifying these two classes. Also, in order to exploit the dependencies of the states of the primary user in time, a Hidden Markov Model is used to improve the performance of the centralized spectrum sensing. Furthermore, a new decentralized cooperative spectrum sensing algorithm is proposed. In this case, the local information of secondary users are appropriately combined to guarantee a reliable communication. Our simulation results indicate the remarkable performance of the proposed cooperative sensing algorithms even in very low signal to noise ratios.