Eigenvalues based spectrum sensing against untrusted users in cognitive radio networks

Spectrum sensing is an essential mechanism for a cognitive radio system. However, the security aspects of spectrum sensing receive little attention so far. In this paper, we identify two kinds of untrusted secondary users which are called dasiaAlways Yespsila users and dasiaAlways Nopsila users. These untrusted secondary users can degrade detection performance greatly, especially when conventional data fusion rules are applied. To counter these threats, for the correlated primary signals, an eigenvalues based detection scheme with double thresholds and revised data fusion rules is proposed. Maximum eigenvalues are proved to be very effective to detect the correlated primary signals and to find the untrusted users. By using the revised data fusion rules, simulation shows that our method has a better detection performance than the conventional method.