Time‐based multi‐component irregular FM micro‐Doppler signals decomposition via STVMD

Abstract

The decomposition of multi-component micro-Doppler signals overlapping in the time-frequency (T-F) domain is critical and challenging, especially in the case of irregular instantaneous Doppler frequency. The authors propose a novel time-based decomposition method called the short-time variational mode decomposition (STVMD) to analyse the irregular (FM) micro-Doppler signals, and present an optimal model combined with T-F transformation. Then, considering the STVMD may fail to extract the instantaneous frequency (IF) of overlapped components, an improved STVMD algorithm is put forward. Since the dependence of the STVMD algorithm on the initial value, they adopt the Kalman filtering to implement IF tracking and regrouping under the global constraint, further accelerating the convergence of the algorithm. Furthermore, due to the mode aliasing at the intersection point, they adopt a degenerate STVMD model to decompose the signals with known centre frequencies, which can be viewed as a Wiener filter. With the two steps, the improved STVMD algorithm can effectively solve the decomposition of T-F overlapping irregular FM micro-Doppler signals. Compared with the peak ridge technique and the ridge path regrouping and intrinsic chirp component decomposition (RPRG + ICCD), the proposed method shows the effectiveness and adaptability even for irregular FM signals with large T-F spectrum amplitude fluctuation in the low signal-to-noise ratio environment.