Target recognition using late-time returns from ultra-wideband, short-pulse radar

Algorithms were developed to recognize targets based on their late-time transient scattering induced by ultra-wideband, short-pulse radar. An improved parameter estimation technique was developed that combined features from two variants of the Matrix Pencil Method. The new algorithm was found to have performance superior to both variants when applied to noisy signals. Parameters estimated by the new algorithm were used to generate Kalman filters that were part of a Multiple Model Adaptive Estimation algorithm that recognized targets by analysis of filter residuals. The Multiple Model Adaptive Estimation algorithm was compared to the Extinction Pulse technique. The algorithm produced over 90% correct identifications in the presence of approximately 20 dB more noise than that applied to the Extinction Pulse technique to achieve similar results.