Robust multichannel detection in heterogeneous airborne radar disturbance

This paper presents the performance of several multichannel adaptive processing detection methods, including a model-based approach which exhibits robustness in correlated disturbances ranging from Gaussian to K-distributed with high tailed probability density functions modeled as compound-Gaussian clutter. Specifically, we consider detection in dense signal environments where training data contains multiple discrete signals in the spatial-temporal domain. For this problem, we compare methods featuring robustness to such processes with the recently proposed non-homogeneity detection (NHD) method, a preprocessing approach for training data selection prior to detection algorithm implementation. Issues considered here include robust detection with respect to clutter texture power variations and multiple signal environments, constant false alarm rate (CFAR) performance and efficient estimation with limited training data.