Counteracting malicious users in cognitive radio networks over imperfect reporting channels

Spectrum sensing data falsification (SSDF) attack are serious threats to collaborative spectrum sensing (CSS) of cognitive radio networks (CRNs). In this paper, inspired by EM (Expectation Maximization) method, we propose a scheme to estimate the presences of primary user (PU) and the SUs´ operating point parameters (false alarm and detection probabilities) iteratively. The key features of the proposed scheme is that, by using the estimated SUs´ operating point parameters, the fusion center can estimate the presences of the PU, while the PU´s state information is feedback to enhance the estimation accuracy of SUs´ operating point parameters. Furthermore, our scheme can achieve a powerful capability of eliminating incorrect sensing reports, which can avoid over penalize the honest users who have random errors in reporting channels. The numerical result shows that, the proposed method can achieve higher malicious user detection accuracy than the existing reputation-based schemes, and can thus improve the CSS performance significantly.