Variational time-frequency mode tracking for micro-Doppler signature extraction

Time-frequency (TF) analysis (TFA) is pivotal for characterizing micro-Doppler (MD) features in radar signals. However, existing methods face challenges to processing passive radar target echoes, such as blurred time-frequency representations (TFRs) and difficulties in extracting instantaneous frequencies (IFs), hindering accurate MD feature description. To overcome these limitations, this paper proposes the variational TF mode tracking decomposition (VTFMTD) method. VTFMTD integrates variational optimization with short-time Fourier transform (STFT)-based analysis to achieve effective mode decomposition and precise IF estimation. The approach comprises two key steps: Decomposing a composite signal into intrinsic modes via TF Wiener filtering, which minimizes the spectral second-order central moment while enhancing reconstruction constraints; and tracking high-fidelity IFs through iterative centroid refinement and smoothing. The method’s effectiveness is validated through simulations and the analysis of actual drone MD signals (MDSs), demonstrating its capability to extract precise micro-motion features. This advancement offers an effective solution for passive radar-based drone surveillance.