CFAR Feature Plane: a Novel Framework for the Analysis and Design of Radar Detectors

TL;DR

This paper introduces the CFAR feature plane framework for analyzing and designing radar detectors, enabling new detector behaviors with high selectivity and robustness without sacrificing detection power or CFAR properties.

Contribution

The paper presents a novel CFAR feature plane framework that allows for the analysis and design of radar detectors with enhanced selectivity and robustness, including new detector types.

Findings

New detectors with diversified behaviors are proposed.

The framework achieves high selectivity without losing detection power.

It provides analytical insights into detector trajectories and shapes.

Abstract

Since Kelly's pioneering work on GLRT-based adaptive detection, many solutions have been proposed to enhance either selectivity or robustness of radar detectors to mismatched signals. In this paper such a problem is addressed in a different space, called CFAR feature plane and given by a suitable maximal invariant, where observed data are mapped to clusters that can be analytically described. The characterization of the trajectories and shapes of such clusters is provided and exploited for both analysis and design purposes, also shedding new light on the behavior of several well-known detectors. Novel linear and non-linear detectors are proposed with diversified robust or selective behaviors, showing that through the proposed framework it is not only possible to achieve the same performance of well-known receivers obtained by a radically different design approach (namely GLRT), but also to devise detectors with unprecedented behaviors: in particular, our results show that the highest standard of selectivity can be achieved without sacrifying neither detection power under matched conditions nor CFAR property.