Decentralized cooperative spectrum sensing in cognitive radio without fusion centre

The cooperative spectrum sensing in cognitive radio networks (CRN) is generally modelled as a distributed binary hypothesis testing problem. Existing literatures mostly deal with cooperative spectrum sensing where a central unit called fusion centre (FC) collects decisions from all secondary users (SUs) and takes the final decision. In this paper, we address the cooperative spectrum sensing scheme without the presence of FC. However, SUs can share their information with each other to make the decision process more reliable. Moreover, modelling of the system in presence of fading is an important issue. The decentralized cooperative spectrum sensing (DCSS) scheme is represented as probabilistic graphical model using factor graph and sum-product algorithm. The new DCSS scheme is proposed by employing likelihood ratio test (LRT) based decision fusion rule. It leads to Neyman-Pearson (N-P) criteria based optimal sensing. Convergence of global decision is also validated using both geographic and broadcast based gossip algorithm for communicating among SUs. Exhaustive simulation results are provided to validate the concept.