Two-sided automatic censoring CFAR detector based on the interquartile range in a heterogeneous Weibull clutter

Abstract

As it is well known, CFAR (Constant False Alarm Rate) detectors are intended to regulate the PFA (Probability of False Alarm) in the presence of a clutter background whose mean power is unknown and/or varying. In heterogeneous clutter backgrounds, CFAR detectors false alarm control and detection efficiency rely upon their ability of clutter edge resilience and/or interfering targets rejection. Aiming to improve radar signal detection in a heterogeneous Weibull clutter, the two-sided automatic censoring IQR-CFAR (InterQuartile Range-CFAR) detector is introduced in this paper. This new detector first resorts to a logarithmic amplifier to dynamically achieve a Weibull to Gumbel transformation of the reference samples. Then, working on a cell-by-cell basis, the CFCR (Constant False Censoring Rate) algorithm retains and rejects samples based on the IQR within the reference window. Upon removal of the unwanted samples, the remaining data set is utilized to estimate the unknown biparametric linear detection threshold through the BLU (Best Linear Unbiased) estimates of the Gumbel distribution’s location and scale parameters. Extensive MC (Monte Carlo) simulations with both simulated and real data show that the suggested detector outperforms its challenging automatic censoring CFAR detectors found in the literature.