Publications

This paper introduces a direct solution of the frequency-dependent, GTD-based, scatterer-model parameters leading towards a new modern spectral-estimation technique to be used for enhanced, super-resolution 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 discretemodel- based analysis eliminates the need for computationally intensive and potentially nonconvergent 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.