Fuzzy C-Means clustering based secure fusion strategy in collaborative spectrum sensing

Collaborative spectrum sensing in Cognitive Radio Networks is vulnerable to Spectrum Sensing Data Falsification (SSDF) attack. Many existing SSDF attack defending algorithms need the assistance of reference sensing devices with high spectrum sensing performance, which are difficult and costly to be implemented in reality. Most of other existing approaches are based on statistic characteristics of sensing information. These approaches assume there is a small fraction of malicious users, and the flip rates of malicious users are fixed and identical. However, some effective attacks such as Sybil attack can be easily launched with which large number of different malicious identities can be generated. To solve these problems, we propose a Fuzzy C-Means clustering based secure fusion strategy. Simulations show this secure fusion strategy outperforms the traditional reputation based approach, at almost any malicious user ratio and varying attack flip rate.