Adaptive decomposition in electromagnetics

Abstract

Decomposition of radar signals into a physically-meaningful signal model is a key step in diagnostic radar imaging, automatic radar target identification, and advanced RCS data post-processing techniques. We review adaptive methods for decomposing radar signals using a very general "dictionary" of functions which can represent disparate scattering mechanisms and are chosen to closely model anticipated characteristics of the signal. The goal of adaptive decomposition is to obtain fast and stable signal representations that are sparse and offer separation of disparate phenomena. The signal decomposition methods we primarily consider are the matching pursuit technique of Mallet and Zhang (1993) and the basis pursuit procedure of Chen and Donoho (1995). We show the matching pursuit decomposition of numerically-simulated data to illustrate that disparate scattering mechanisms can be isolated and identified.