Performance of adaptive optimal Doppler processors in heterogeneous clutter

The performance of adaptive Hsiao-optimal clutter-suppression filters operating in spatially varying clutter backgrounds is investigated. The case of practical adaptive estimation of the covariance matrix, and limiting cases where the estimate of the covariance matrix tends to the mean clutter covariance matrix, are considered. Three classes of nonhomogeneous clutter are studied: the amplitude and spectral width in each range bin are drawn from spatially invariant independent gamma distributed parent populations; clutter edges, in which the range profiles of the amplitude and/or spectrum exhibit a step change; and homogeneous clutter in which some range bins are corrupted by point-clutter sources or extraneous targets with significantly different amplitude and spectral characteristics. The use of prefilter MTI is investigated as a means of reducing losses in improvement factor (IF) in some classes of clutter nonhomogeneity. It is shown that adaptive filters suffer significant reductions in IF in nonhomogeneous clutter environments. These losses are generally greater than the benefit afforded by ideal adaptive processors over other conventional Doppler processors. The mean IF loss is an optimistic performance measure in adaptive Doppler processors