A new CFAR detector based on PDF estimation of clutter using maximum entropy method

Abstract

A new CFAR (constant false alarm rate) algorithm is proposed for the Blind-CFAR detector which is proposed for CFAR detection in various types of clutter. The new algorithm employs the maximum entropy (MaxEnt) method to estimate the PDF (probability density function) of clutter and thus the CFAR detection thresholds based on the samples of reference cells and some moment constraints. Performance of the new algorithm is proved to converge to the performance of the ideal detector as the numbers of the reference cells and the moment constraints approach to infinity. With finite reference cells and moment constraints, performance of the new algorithm is compared with that of the maximum likelihood (ML) CFAR detector in uniform Weibull clutter with known and mismatched shape parameters. The results show that, the Blind-CFAR detector employing the new algorithm performs as well as the ML-CFAR detector in case of known shape parameters and it outperforms the ML-CFAR detector in case of mismatched shape parameters.