Less Sample-Cooperative Spectrum Sensing in the Presence of Large-Scale Byzantine Attack

Cooperative spectrum sensing (CSS) has emerged as a promising solution to identify available spectrum resources in cognitive wireless sensor networks (CWSNs). However, such an open collaboration paradigm may suffer Byzantine attack from malicious sensors, leading to a dramatic decline in cooperative performance, meanwhile, in a large CWSN, the CSS process causes significant communication overhead in the common control channel to report local measurements, which limits the cooperative efficiency. On this account, a less sample-CSS, weighted sequential detection (WSD), is proposed to prevent large-scale Byzantine attack. First, the channel state information is used to verify the robustness of the global decision made by the fusion center and evaluate the trust value of sensors. Furthermore, the weight of each sensing sample related to trust value is integrated into a sequential detection within a time window. Finally, a sequential approach in a trust value descending order is formulated. A series of numerical simulation results show that compared with alternative fusion rules for CSS, the advantages of the proposed WSD in terms of the error probability and the average number of samples are evident in the presence of a large-scale Byzantine attack.