Multi-Target Detection Using Early-Time Response Based on the Joint of Half Fourier Transform and Short-Time Matrix Pencil Method

The detection and identification technology based on target wideband scattering has become one of the effective solutions currently available for radar target detection, holding promising application prospects in fields such as the Internet of Vehicles (IoV), military affairs, and biomedical engineering. The proposed method in this paper addresses the issues of missing late-time response information in far-field target echo and target detection omissions caused by multi-target signals aliasing. By analyzing the early-time response characteristics of targets, we present a joint algorithm combining half Fourier transform and short-time matrix pencil method (HFT-STMPM) for radar target detection and localization. The short-time matrix pencil method (STMPM) enables target detection by extracting the early-time characteristics of signal. However, since the extracted early-time characteristics are often contaminated by late-time response, preprocessing in the half Fourier transform (HFT) domain is required to differentiate and purify multiple early-time responses embedded in echo signal. This approach can significantly improve computational efficiency and target detection accuracy in subsequent processing stages. In HFT-STMPM, to enhance pole extraction accuracy, we optimize three key parameters: bundle parameter, mode number, and window length. Through analyzing the distribution characteristics of early-time response poles, a piecewise smooth curve fitting method is proposed to improve computational precision for determining the center moment of early-time response. The results demonstrate that HFT-STMPM can accurately distinguish and detect multiple targets while achieving improved positioning accuracy among them. Furthermore, this method shows potential contributions in characterizing electromagnetic properties of complex targets.