Malicious users detection in collaborative spectrum sensing using statistical tests

Collaboration among cognitive radios has been extensively studied in the past and widely accepted as a viable approach to improve spectrum sensing reliability. Data fusion in collaborative spectrum sensing rely on the information received from cognitive radios. It has been shown in literature that the performance of collaborative spectrum sensing degrades significantly in the presence of even a single malicious user. In this paper, a new scheme to detect and eliminate malicious users in collaborative spectrum sensing is proposed. Our method is based on the Grubb's test and is able to detect and eliminate observations of multiple malicious users. Simulation results show that the proposed scheme has much higher detection probability in the presence of malicious users especially for the case when the received signal to noise ratio is low.