Model-based clutter cancellation based on enhanced knowledge-aided parametric covariance estimation

Aerospace radar operation requires robust target detection in the presence of strong ground clutter returns. Space-time adaptive processing (STAP) is a leading approach to cull weak target signals from a strong clutter background. A scene comprised of heterogeneous clutter degrades typical STAP implementation by corrupting training data, thereby leading to covariance estimation error, corresponding adaptive clutter filter mismatch, and loss of detection performance. In the knowledge-aided parametric covariance estimation (KAPE) method, the processor estimates the parameters of a validated clutter covariance model to tailor the clutter filter response to rapid changes in heterogeneous clutter environments. Computational burden and array manifold estimation, given antenna complex gain errors, are two primary concerns when implementing KAPE. This paper develops and characterizes an enhanced KAPE (E-KAPE) approach. Specifically, we develop a computationally efficient implementation via application of Gram-Schmidt orthonormalization to the modeled clutter manifold and discuss an iterative approach to estimate complex gain errors from channel-to-channel, leading to substantial improvement in model-based clutter cancellation potential. For the specific simulation results shown, we find a 9-dB improvement in signal-to-interference-plus-noise ratio loss for our proposed approach relative to the original KAPE method. Furthermore, for these cases, we show that our E-KAPE method outperforms conventional STAP algorithms in the presence of homogeneous and heterogeneous clutter.