Performance of two Low-Rank STAP Filters in a Heterogeneous Noise

This paper considers the Space Time Adaptive Processing (STAP) problem where the disturbance is modeled as the sum of a Low-Rank (LR) Spherically Invariant Random Vector (SIRV) clutter and a zero-mean white Gaussian noise. To derive our adaptive LR-STAP filters, the estimation of the projector onto the clutter subspace is performed from the Sample Covariance Matrix (SCM) and the Normalized Sample Covari-ance Matrix (NSCM). We compute the theoretical performance of both corresponding LR-STAP filters through the analysis of the Signal to Interference plus Noise Ratio (SINR) Loss, based on a perturbation analysis. Numerical simulations validate the theoretical formula and allow to show that the LR-STAP filter built from the SCM performance does not depend on the heterogeneity of the SIRV clutter whereas the LR-STAP filter built from the NSCM performance does.