Speaker: Geetha Chandrasekaran (13So41)
Energy detection (ED) is a simple method for Cognitive Radio spectrum sensing when the characteristics of primary user signal are unknown. In this work, we attempt to understand the performance of energy detector in the presence of generalized channel fading. In the process, we have derived the expression for the Probability of detection under k-u shadowed fading, which has provided a new insight into the impact of detector’s threshold on the detection probability. It has been proven analytically that the probability of detection is a convex function of the detector threshold. We also analytically characterize the latency involved in performing Cooperative Spectrum Sensing (CSS) over a static spectrum sensing cycle (SSC) and show that improved detection performance is at the cost of network latency. Hence, we propose a dynamic SSC for CSS which strikes a trade-off between improved detector performance and latency. Furthermore, analysis of ED’s performance in a more realistic scenario (where the arrival/ departure time of the primary user is unknown) has been done for the simple Gaussian Mixture model. We find that using Robust statistics provides an improvement in the performance of ED as compared to the classic ED case.