Power-Law Scattering Models and Nonlinear Parametric Estimation for Super-Resolution Radar

摘要

This paper introduces a direct solution of the frequency-dependent, GTD-based, scatterer model parameters leading towards a new modem spectral estimation technique to be used for enhanced superresolution RADAR analysis. The overcomplete nature of the full GTD scatterer model basis (positive and negative half-integer power laws) is recognized and overcome by introducing the vector channel method, well known from communication theory. This physically motivated discrete-model based analysis eliminates the need for computationally intensive and potentially non-convergent local optimization procedures. Each scatterer is assigned a half-integer power law that identifies its cross section frequency dependence and hence restricts the possible underlying physical feature geometries. This analysis opens the possibility for vector-attribute based feature processing for target recognition that offers the potential for significant improvement in target identification performance.