Optimal Generic L-Out-of-M Counting Rule for Neyman-Pearson Test in Cognitive Radio Networks

This letter proposes a generic ${L}$ -out-of- ${M}$ counting rule-based sensing, wherein the cognitive radios (CRs) with non-identical receiver operating characteristic (ROC) curves and the fusion node cooperatively identify the primary user’s state. We formulate a generalized Neyman-Pearson problem to jointly optimize the individual CRs’ operational points on the ROC curves and the generic ${L}$ -out-of- ${M}$ counting rule for the CR system. Then, a fast-sensing problem is formulated to find the least number of CRs needed for practical sensing. We provide generalized solutions for any detector employed in the CR system. The proposed scheme shows superior detection performance compared to the traditional scheme.