Dynamic suspicious reputation based collaborative sensing in cognitive radio networks

Collaborative sensing is an available technology to increase access opportunity for unused spectrum and avoid interference to the primary users in cognitive radio networks. Due to the complexity of wireless environment and the primary user's location is unknown, the sensing performance will be affected by many unfavorable factors such as multipath fading, shadowing, and so on. Especially, some secondary users (SUs), defined as attackers, could cause security problem by reporting false information to other SUs. In this paper, we proposed a Dynamic Suspicious Reputation Scheme to detect attackers in collaborative sensing. In the proposed scheme, a SU with suspicious reputation makes the sensing decision by evaluating the correlation value between its own sensing reports and those from other SUs. If the correlation value exceed a given threshold, the compared SU is regarded as an attacker. The suspicious reputation is updated according to its own sensing reports in each collaborative sensing iteration. Once the suspicious reputation of a SU exceed a certain threshold, the SU is regarded as Malicious Attacker, and removed out the collaborative sensing group. Numerical analysis show that the proposed scheme can achieve a considerable performance improvement compared with a User-centric Misbehavior Detection Scheme for secure collaborative sensing. In addition, we also discuss the affection of the suspicious reputation to the probability of false alarm and detection.