Sample-Starved Wavefront Adaptive Sensing and GLRT for MTI Radar

Moving target indicator (MTI) radars can suffer signal-to-noise ratio (SNR) losses due to: 1) the use of heavy non-adaptive tapers, and/or 2) poor estimation of space-time adaptive weight vectors. For ground-based radars, non-adaptive temporal and spatial tapers with low slide lobe levels are required to suppress strong ground clutter near zero-Doppler. For space-time adaptive processing (STAP), target-free training data is often limited by clutter inhomogeneity in range. In this paper, blind source separation (BSS) which exploits the different Doppler spectra of targets versus clutter is employed to obtain target-free training data using only the space-time data from each range cell-under-test. Two methods using BSS outputs are presented: 1) a wavefront adaptive sensing (WAS) beamformer for suppressing clutter with subsequent matched filtering in the Doppler domain, and 2) a clutter-subspace based optimal generalized likelihood ratio test (GLRT) detection statistic. WAS is used here with non-recurrent (e.g. multi-pulse-frequency-repetition) waveforms to permit detection of targets otherwise masked by Doppler-aliased clutter. Simulation results for a 2-D ground-based S-band phased array MTI radar are presented which demonstrate significant SNR gain versus conventional methods.