Performance of cooperative spectrum sensing with correlated cognitive users' decisions

Abstract

Cooperative spectrum sensing is employed in cognitive radio network to reliably detect the primary users' transmissions by fusing the sensing data of individual secondary users. In this paper, we study the performance of cooperative spectrum sensing, in terms of the system probability of detection, when the secondary users' local decisions are correlated. We use a correlation model that is indexed by a single parameter and fix the fusion rule to one of three decision rules which are the OR, AND and Majority Voting rules. Our results show that the performance of cooperative spectrum sensing degrades with the increase in correlation between the secondary observations for all the fusion rules considered. We also show that, whether the OR or Majority Voting rule is superior depends mainly on the correlation index. When the secondary users' local decisions are independent, the Majority Voting rule outperforms the OR and AND fusion rules. However, as the correlation between the local decisions increases, the OR fusion rule outperforms the other two rules. Also, as the correlation index increases, for the same system probability of false alarm, higher signal-to-noise ratio is required to be received at the secondary users to achieve the same system probability of detection for all the fusion rules considered.