Robust non parametric CFAR detector in compound Gaussian clutter

One of the most important problems in radar detection design is the maintaining of Constant False Alarm (CFAR). In the context of non coherent detection for high resolution radars with low grazing angle, we propose a new decision rule that achieves the CFAR property with respect to a class of clutter models. This paper demonstrates that the new proposed detector is CFAR and robust under the assumption of the compound Generalized Pareto (GP), the K distribution and the compound inverse Gaussian distribution. The results obtained are examined and discussed.