Robust Radar Detection of a Mismatched Steering Vector Embedded in Compound Gaussian Clutter

TL;DR

This paper introduces a robust radar detection method for scenarios with mismatched steering vectors in compound Gaussian clutter, using semi-definite programming to improve detection performance under mismatch conditions.

Contribution

A novel robust detection technique based on the generalized likelihood ratio test that handles steering vector mismatches in compound Gaussian clutter environments.

Findings

Detector outperforms in mismatched scenarios

Negligible loss in perfectly matched cases

Maintains constant false alarm rate

Abstract

The problem of radar detection in compound Gaussian clutter when a radar signature is not completely known has not been considered yet and is addressed in this paper. We proposed a robust technique to detect, based on the generalized likelihood ratio test, a point-like target embedded in compound Gaussian clutter. Employing an array of antennas, we assume that the actual steering vector departs from the nominal one, but lies in a known interval. The detection is then secured by employing a semi-definite programming. It is confirmed via simulation that the proposed detector experiences a negligible detection loss compared to an adaptive normalized matched filter in a perfectly matched case, but outperforms in cases of mismatched signal. Remarkably, the proposed detector possesses constant false alarm rate with respect to the clutter covariance matrix.