A Locally Optimal Soft Linear-Quadratic Scheme for CR Systems in Shadowing Environments

Abstract

In this letter, we analyze the problem of detecting spectrum holes in cognitive radio systems under the Neyman-Pearson scenario. We consider that a group of unlicensed users use noncoherent energy detectors to sense the radio signal and design a soft locally optimal linear-quadratic statistic based on the deflection coefficient. Each unlicensed user transmits the processed data to a central entity, where the decision about the presence or not of licensed users is made. Using the method of Monte Carlo, we show that the proposed statistic outperforms previous ones available in the literature in a wide range of shadow-fading scenarios and it is robust against parameters errors.