Fast Iterative Subspace Algorithms for Airborne STAP Radar

Hocine Belkacemi,Sylvie Marcos

doi:10.1155/asp/2006/37296

Abstract

Space-time adaptive processing (STAP) is a crucial technique for the new generation airborne radar for Doppler spread compensation caused by the platform motion. We here propose to apply range cell snapshots-based recursive algorithms in order to reduce the computational complexity of the conventional STAP algorithms and to deal with a possible nonhomogeneity of the data samples. Subspace tracking algorithms as PAST, PASTd, OPAST, and more recently the fast approximate power iteration (FAPI) algorithm, which are time-based recursive algorithms initially introduced in spectral analysis, array processing, are good candidates. In this paper, we more precisely investigate the performance of FAPI for interference suppression in STAP radar. Extensive simulations demonstrate the outperformance of FAPI algorithm over other subspace trackers of similar computational complexity. We demonstrate also its effectiveness using measured data from the multichannel radar measurements (MCARM) program.

Highlights

Space-time adaptive processing (STAP) is a technique for suppressing clutter and jamming in airborne radar [1]
To reduce the computational burden linked to the singular value decomposition (SVD), many algorithms have been proposed in spectral analysis and spatial array processing literature
We more precisely propose the application of the fast approximate power iteration algorithm (FAPI) [6] in order to deal with the interference suppression in STAP radar

Summary

Introduction

Space-time adaptive processing (STAP) is a technique for suppressing clutter and jamming in airborne radar [1]. A common method to obtain these subspaces is via singular value decomposition (SVD) of the interference-plus-noise covariance matrix. Such methods can reduce the sample support requirement to O(2r), where r is the rank of the covariance matrix, but at the expense of a considerable computational complexity due to the SVD O((NM)). Such methods can reduce the sample support requirement to O(2r), where r is the rank of the covariance matrix, but at the expense of a considerable computational complexity due to the SVD O((NM)3)

Objectives

Results

Conclusion