Performance evaluation of RQ non‐parametric CFAR detector in multiple target and non‐uniform clutter

Abstract

The performance analysis of the rank quantisation (RQ) non-parametric constant false alarm rate (CFAR) detector in homogeneous and non-homogeneous backgrounds is carried out. A closed-form expression of the false alarm rate of the RQ detector at clutter edges is derived. It is shown that the RQ detector pays an additional loss of about 3 dB in detection performance relative to the cell averaging (CA)-CFAR with incoherent integration in homogeneous background, but it can accommodate a number of strong interfering targets within the reference window. When both of the heavy and the weak clutters follow a Rayleigh distribution, the rise of false alarm rate for the RQ detector at clutter edges lies between that of the CA-CFAR and the ordered statistic (OS)-CFAR with incoherent integration. If the Weibull distributed clutter with a long tail moves into the reference window, the rise of the false alarm rate of the CA-CFAR, the OS-CFAR and the greatest-of-CFAR with incoherent integration reaches a peak of more than three orders of magnitude, and cannot return to the original design P fa. However, the RQ detector reveals its inherent virtue that it can maintain a CFAR despite the changes in the underlying distributions of observed data.