Late-Time Estimation Methods to Achieve Resonance Based Radar Target Recognition

The radar target scattered response can be given by early and late-time response based on the singularity expansion method (SEM). The late-time provides information about the target geometry and is theoretically independent of aspect angle. The SEM theory defines the late-time response as a combination of the unique complex natural resonant pole frequency. Thus, it is important to estimate the late-time index correctly and extract the resonant poles. This paper investigates the late-time estimation algorithms such as the Auto-Regressive (AR) method, Hankel matrix-based approach, and a recent Short-Time Matrix Pencil Method (ST-MPM) approach. The main contribution of this paper is to provide a summary of all existing approaches. We conducted the simulation analysis for four canonical objects by considering the additive white Gaussian noise. Moreover, to check the accuracy of each method based on a quantitative measure of Root Mean Square Error (RMSE) between reconstructed and original late-time signal.