Performance comparison of three recent STAP methods

In this paper, we present a comparison of target detection performance for normalized adaptive matched filter (NAMF), normalized parametric adaptive matched filter (NPAMF), and normalized low-rank adaptive matched filter (LRNAMF) for space-time adaptive processing. Test statistics for these algorithms as functions of range bins and filter outputs as functions of Doppler beam position (DBF) and azimuth angle (AZ) are computed for the KASSPER L-band datacube, which is simulated for the airborne linear phased array radar application. First, we illustrate that when the signal-to-noise ratio is used as a target-detection parameter, LRNAMF outperforms NAMF and NPAMF under weak conditions of training data contamination. Next, we demonstrate the target cancellation effect when the training data are contaminated by competing targets (outliers). Finally, we present a scenario for target detection in heterogeneous radar clutter when there is spatio-temporal steering vector uncertainty. In this scenario, we show that there is substantial broadening in the filter outputs as functions of DBP and AZ for these algorithms.