Clutter Subspace-Based Space-Time Adaptive Radar Detectors

A new class of adaptive radar target detection tests is considered that is based on a data model that models the clutter in the mean of the distribution. This mean characterizes the space-time subspace of the clutter and is motivated in part by a channel matrix-based approach, but rather focuses on modeling the data after pulse compression (which yields a simpler model). This assumed data model is a departure from the classical model for pulse-compressed radar data that models the clutter stochastically via the data space-time covariance matrix. The new detection tests obtained are more efficient in terms of required sample support, and remain viable even in the absence of secondary target free training data. This class includes new versions of the generalized likelihood ratio test, the adaptive matched filter, the adaptive coherence estimator, and the 2-D adaptive sidelobe blanker.