Adaptive temporal processing for spread Doppler clutter mitigation in over-the-horizon radar

The detection performance of a sky-wave, HF over-the-horizon radar (OTHR) essentially depends on its ability to distinguish a weak target from strong surface back-scatter clutter. Doppler processing is the conventional technique that discriminates moving targets from nominally stationary clutter. However, disturbances of the radar's transmission medium, the ionosphere, sometimes spreads clutter in Doppler space, thereby obscuring weak targets. This work investigates methods for mitigating spread-Doppler clutter (SDC). Specifically, SDC occurring during day-night transitions in equatorial regions is addressed by considering the spatial correlation of moving irregularities due to the earth's magnetic field lines. For a radar looking south from the north, aberration in radar return should be highly correlated within a neighborhood of range bins. Adaptive temporal processing aims to take advantage of this property to estimate the SDC covariance with sample support in range. The covariance of SDC is in turn used in adaptive cancellation of clutter with adaptive matched filtering (AMF). Inversion of a covariance matrix estimate with small sample support requires a regularization technique. Eigenvalue thresholding is chosen for regularization. In this thesis, eigenvalue thresholding is shown to provide a maximum likelihood (ML) estimate from the sample covariance matrix, given that the white noise variance in data is known a priori. In the context of source localization, the relation between diagonal loading and eigenvalue thresholding is studied. In addition, the relationship between ML source localization and minimum variance (MV) beamforming is established by the judicious choice of a signal model. Real data with realistic target injection is used in simulations, which indicate that a gain in detection performance in the order of 10 dB SNR is achievable with the AMF.