Knowledge-Aided Direct Data Domain STAP Algorithm for Forward-looking Airborne Radar

Abstract

A knowledge-aided direct data domain (D3) algorithm, which is applied to the space-time adaptive processing (STAP), is proposed. The conventional D3 algorithm only uses the test data to avoid heterogeneous training data. However, the prior knowledge of target is always not accurate, leading to performance degradation. The D3 algorithm based on sparse representation suffers target self-nulling and clutter suppression performance degradation. To overcome these disadvantages, a knowledge-aided D3 algorithm is proposed to improve the STAP performance. The prior knowledge about clutter distribution of forward-looking airborne radar is exploited to distinguish clutter components and targets. Furthermore, the clutter components can be eliminated from the sparse recovery coefficient vector. Hence, we can judge if the targets are present via the power of reminder element. Then, the clutter components are exploited to estimate the STAP filter weight, which can supply satisfactory clutter suppression performance. Experiments validate that the proposed algorithm separate the clutter components and targets directly without the prior knowledge of targets, and the target detection performance is improved significantly.