Separation of Multi-component Micro-Doppler Signals Based on HHT-GAMD

Abstract

Radar micro-Doppler is the phenomenon of the observed micro-motions on top of the bulk main Doppler component of a target's motion. Joint time-frequency analysis is the tool of first choice for micro-Doppler features extraction, among which Hilbert-Huang transformation (HHT) is able to extract exact instantaneous micro-Doppler feature by the Hilbert transform. To overcome the shortcoming of HHT in distinguishing the signals with closely frequency components, a new separation method for micro-Doppler signals is proposed, which is based on HHT and generalized analytical mode decomposition (HHT-GAMD). Firstly, empirical mode decomposition (EMD) is applied to decompose the original signal into certain intrinsic mode functions. The frequency spectrum along with frequency values are produced by Hilbert transform. Then, generalized analytical mode decomposition method extracts the signal with different frequency components and separates it to single frequency component based on the judgment whether the considered component contains two or more frequency components. Theoretical studies are displayed and performance considerations are covered with numerical simulations. The results indicate that, compared with HHT method, the proposed method is demonstrated to have efficient decomposition of the micro-Doppler signals and improves the accuracy.