Target Focused Estimation of the Stages in the Multistage Wiener Filter
Rachel Gray, Elias Aboutanios, Josef Zuk, Luke Rosenberg, David Kirszenblat · 2024
Adaptive detection techniques for airborne radar suffer in non-stationary, non-homogeneous environments and when there is limited sample support. The multistage Wiener filter (MWF) is a reduced rank adaptive detector that reduces the required sample support while lowering the computational burden. Key to these benefits is the provision of a reliable method to determine the number of MWF stages. Current approaches estimate the rank of the entire clutter subspace, which often overestimates the number of MWF stages. Consequently, we reformulate the problem and propose an algorithm that determines the number of stages required to achieve the maximum probability of detection. The algorithm is formulated as a model order estimation problem and utilises log-likelihood ratios. We provide a computationally efficient implementation and demonstrate its effectiveness through simulations.