Asymptotically Optimal Likelihood detector for cyclostationary signature induced by Cyclic Delay Diversity

Abstract

The Cyclic Delay Diversity (CDD)-induced cyclostationary signature is considered to be a robust and cost-efficient scheme for self-coordination of Cognitive Radio Network (CRN). However, the performance of network coordination relies on the reliable detection of such cyclostationary signatures. In this paper, we deduce an exact covariance matrix to characterize the statistics of cyclostationary signature. Based on the covariance matrix, we propose an Asymptotically Optimal Likelihood (AOL) detector for the test of the CDD-induced cyclostationary signature. In addition, an Asymptotically Maximum Likelihood Probability (AMLP) criterion is provided to solve the multiple signatures identification issue. Comprehensive simulations verify that the proposed detector provides superior performance in detection probability and observation duration, compared with the existing Constant False Alarm Rate (CFAR) detector.