High-Accuracy Cardiac Activity Extraction Using RLMD-Based Frequency Envelogram in FMCW Radar Systems

This paper proposes an accurate extraction of beat-to-beat interval (BBI) using the robust local mean decomposition (RLMD)-based frequency envelogram (FEnv) method in frequency-modulated continuous wave (FMCW) radar systems with lower signal-to-noise ratio (SNR). The FEnv benefits from spectral integration of time-frequency representation (TFR) and can further emphasize the vital signal fragments with strong characteristics. However, the performance of original FEnv suffers from the insufficient time-frequency resolution of the Fourier transform, resulting in poor BBI estimates. RLMD decomposes the reflected signals into the amplitude-modulated (AM) signals and the frequency-modulated (FM) signals, which can be reshaped according to heartbeat frequency band, to improve time-frequency resolution. After proper preprocessing of the received I/Q signals, the cardiac activity can be extracted accurately by the proposed RLMD-based FEnv. The results of the proposed algorithm are in high agreement with the reference signals ECG, and the error rate can reach 1.84% on average.