Research on the accuracy of Gaussian approximated expressions for energy detection in cognitive radio networks

This paper addresses the problem of the adaptable conditions for Gaussian approximated expressions of energy detection over AWGN and Rayleigh channels. Gaussian approximated expressions of false alarm and detection probabilities have been widely adopted for evaluating the sensing performance of energy detection when the number of samples is large. With deeper study, we find the fact is not such. Extensive analysis and numerical simulations confirm that the approximated expressions can lead to significant errors in the performance of cooperative spectrum sensing over AWGN and Rayleigh channels, especially for a large cooperative network. Also, the approximated expressions are not applicable to the local detection over AWGN and Rayleigh channels, when a lower false alarm probability is required. Moreover, Gaussian approximated expressions overestimate the detection performance in all cases.