Performance Evaluation of Energy Detection for Correlated Random Signals

This paper investigates the performance of energy detection (ED) for random signals having sufficient autocorrelation. As opposed to the uncorrelated noise, practical communication signals have autocorrelation, which can be exploited for optimal detection if the correlation is known a priori. Otherwise ED can be employed as a generalized sensing method for unknown signals. However, since ED is optimal only when signal samples are uncorrelated, its performance degrades due to the correlation present. We investigate the detection performance of ED both analytically and by simulation under the assumption of Gaussian signal having exponential autocorrelation function. Analysis and simulation show that the performance degradation is negligible for low signal-to-noise ratio (SNR) levels, and also for the first-lag correlation less than 0.5. For highly correlated signals, the degradation is pronounced when SNR is above -12 dB.