Maximum likelihood estimation of clutter subspace in non homogeneous noise context

Abstract

In the context of a disturbance composed of a Low Rank (LR) clutter plus a white Gaussian noise, the corresponding LR filters used to detect a target embedded in this disturbance needs less training vectors than classical methods to reach equivalent performance. Unlike the classical one which is based on covariance matrix of the noise, the LR filter is based on the clutter subspace projector. In this paper, we propose a new estimator of the clutter subspace projector for a disturbance composed of a LR Spherically Invariant Random Vectors (SIRV) plus a zero mean white Gaussian noise that does not require prior information on the SIRV's texture. Numerical simulations validate the introduced estimator, and its performance and robustness are tested on a Space Time Adaptive Processing (STAP) simulation.