Adaptive CFAR detection via Bayesian hierarchical model based parameter estimation

Radar CFAR detection is addressed in this paper where the unknown noise/clutter statistics are modeled using a hierarchical structure. Considering the secondary data as a probability mixture due to the complex and heterogeneous background, parameter estimation is achieved using the empirical Bayesian approach. Unlike conventional cell averaging CFAR (and its variations) and order statistics CFAR, the new CFAR detection algorithm is less sensitive to the clutter edge location/duration. Performance evaluation is conducted via numerical simulation.