New Approach in Time-Frequency Analysis with Applications to Experimental High Range Resolution Radar Data

摘要

This report presents trade-off studies on Time-Frequency Distribution (TFD) algorithms and a methodology for fusing them to achieve better target characterization. It is shown that TFD algorithmic fusion considerably increases the detectability of signals while suppressing artifacts and noise. The report reviews a sample of representative TFD algorithms. Their performance is studied from a qualitative and quantitative point of view. For simplicity, we considered the mean-squared error as a measure of performance in the quantitative trade-off studies. The TFD algorithmic fusion is performed using a self-adaptive signal. It may be adjusted to work for a wide range of signal-to-noise ratios. The algorithm uses the first two terms of the Volterra series expansion and we treat the outputs of the time-frequency algorithms as the variables of a Volterra series and the coefficients of the series are estimated through training sets with a least-squares algorithm. Simplistic TFD algorithmic fusion methods (e.g., weighted averaging or weighted multiplication) are special cases of the proposed fusion technique. We demonstrate the effectiveness of TFD algorithmic fusion method using experimental High Range Resolution (HRR) radar data and simulated data.