Radar Clutter Modeling Based on CGIG and Mixture CGIG Distributions

Development of target CFAR (Constant False Alarm Rate) detection schemes requires the specification of radar clutter models. There is why the modeling study of non-stationary radar sea-clutter statistics is a first serious research topic in favor of the above challenge. Several scenes of real data consist of mixture Gaussian and non-Gaussian statistics. To obtain an accurate fitting to IPIX (Intelligent Pixel X-band) sea clutter, compound Gaussian Inverse Gaussian (CGIG) class distribution is considered in this work against standard distributions labeled Weibull, log-normal, Pareto type II and K. Parameters values of the above models are estimated from IPIX data using MLE (Maximum Likelihood Estimation) method and LSA (Least Squares Approximation) method. In most cases of IPIX database, experimental study shows that CGIG and mixture CGIG CCDFs (Complementary Cumulative Distributed Function) can fit empirical CCDF in terms of range cell resolution, antennas polarization and range cell number.