Cooperative Sensing With Correlated Local Decisions in Cognitive Radio Networks

In this paper, we analyze the impact of correlated secondary users' local decisions on the performance of cooperative spectrum-sensing schemes when the counting rule is employed at the fusion center. We employ a correlation model that is indexed by a single parameter ρ. We derive the system probabilities of detection and false alarm for the
K
-out-of-
M
counting rule when the secondary users' local decisions are correlated under both hypothesis. Our performance evaluations are based on two performance criteria, which are the Neyman-Pearson (NP) criterion and the minimization of the sensing errors. Our results show that, for each value of the correlation index, there exists an optimal value of
K
that satisfies each criterion. We use genetic algorithm to find the optimal setting that minimizes the total probability of sensing error since the optimization problem under the correlation model used in our analysis is a mixed integer nonlinear problem with nonlinear constraint.