Performance analysis of two modified censored mean-level CFAR detectors in Pearson-distributed reverberation

Abstract

In this paper, the performances of two modified censored mean-level (CM) Constant False Alarm Rate (CFAR) detectors in Pearson-distributed (PD) reverberation are studied. Both of the detectors generate two local estimations with CM and cell averaging (CA) methods according to the leading and lagging reference windows respectively, and take the greater value (CMCAGO) or smaller value (CMCASO) of the two local estimations as the reverberation level estimation. We derive the analytical expressions of the two detectors' false alarm probabilities and detection probabilities in homogeneous PD reverberation, and prove that both of the detectors have the CFAR property with respect to the reverberation level. In strong multi-target interference environment, we analyze and derive the detection probability expressions of both of the detectors. Through simulations, the performances of the two detectors are comprehensively analyzed and compared with traditional CFAR detectors like the OS, SO and GO. The results show that the detection performance of the CMCASO is close to that of the OS, and the CMCAGO has a very stable detection performance in multi-target interference environment.