Knowledge-aided GMTI in a Bayesian framework

Limited or heterogeneous training data restrict the ability to estimate clutter covariance for radar detection of moving targets. In addition, traditional range-cell by range-cell processing neglects the effects of target sidelobe leakage across range, increasing false alarms and limiting the dynamic range of detectable targets. We propose a knowledge-aided Bayesian formulation for moving target detection that admits fast computation. The approach jointly images clutter and targets, learns clutter from a single range, and deconvolves target sidelobes across range. The proposed processing is illustrated on KASSPER I data.