Performance Comparison of Joint Correlation and Improved Energy Detection for Spectrum Sensing

Abstract

Spectrum sensing is a important feature in cognitive radios, wherein the idea is to detect the presence of the incumbent on a licensed spectrum. It is known that the improved energy detector (IED)- which calculates the sample norm of an optimally chosen order in the received observations - outperforms the conventional energy detector (ED). The relatively recently proposed joint correlation and energy detector (CED) combines the sample energy and the first order correlation values in the received observations, and it is known that when the combining is carried out optimally, it outperforms ED. We present a comparative performance study of CED and IED, for different incumbent signal models, following the Neyman-Pearson framework. We show that the CED consistently outperforms IED across the Gaussian, sinusoidal and constant signal models. Similar to what was reported earlier, the IED is observed to outperform ED only for low values of probability of false-alarm. Hence, this study establishes the applicability of CED for spectrum sensing, as opposed to IED and ED.