Reliable Cooperative Sensing in Cognitive Networks

In this paper, we consider reliable cooperative spectrum sensing in cognitive networks under the Spectrum Sensing Data Falsification (SSDF) attacks. One effective method to mitigate the SSDF attacks is the q-out-of-m fusion scheme, where the final decision is based on q sensing reports from m polled users. In this paper, first, we derive the asymptotic behavior of the fusion scheme as the network size increases. It is found that the false alarm rate decreases exponentially as the network size increases, even if the percentage of malicious users remains fixed. Second, we propose an iterative approach to obtain the best scheme parameters that minimizes the false alarm rate and enforces the miss detection constraint. Third, we discuss different attack scenarios and propose a malicious user detection method to further improve the performance. It is shown that by exploiting the malicious user detection scheme, the system performance is improved significantly under various attacks.