Adaptive CFAR Detection for Clutter-Edge Heterogeneity Using Bayesian Inference

Reliable radar constant false alarm rate (CFAR) detection is critical in dynamic clutter and jamming scenarios [2]. The existence of heterogeneities in practical operational environments renders the conventional cell averaging CFAR (CA-CFAR) ineffective. Heterogeneities arise due to the presence of multiple targets and clutter edges [3]. Alternative schemes have been developed to address this issue, including order statistic CFAR (OS-CFAR) and its variations [4–6] as well as various windowing techniques aimed to exclude heterogeneous regions. Nonetheless, each scheme is targeted toward a particular clutter/interfering target scenario. For example, a variation of CA-CFAR, called the greatest of CFAR (GO-CFAR), calculates the average of the leading and lagging windows, respectively, and selects the greater of the two as an estimate of the clutter strength. Clearly the underlying assumption is that the nonhomogeneity (e.g., clutter edge) appears in either the leading or lagging window, but not both. Systematic analysis of various CFAR detectors in nonhomogeneous background is given in [7]. Attempts have also been made to intelligently select