A byzantine attack defender for censoring-enabled cognitive radio networks

This paper considers the problem of cooperative spectrum sensing (CSS) in the censoring-enabled Cognitive Radio Networks (CRNs) with a crowd of battery-powered Secondary Users (SUs), where only significant local observations are submitted to the Fusion Center (FC). However, in order to monopolize the spectrum usage or disrupt the networks operation, malicious SUs may try to send falsified sensing reports to the FC even they are uncertain about their observations, which make the existing robust CSS schemes ineffective. To tackle these challenges, we formulate an optimization problem to improve the performance of CSS in the censoring-enabled CRNs, and develop an expectation maximization based algorithm to solve it, where the presences of primary user and the reliabilities of each SU can be jointly estimated. Extensive simulation results show that the proposed robust CSS scheme outperforms the previous reputation-based approaches under various attack scenarios.