Respiration and Heartbeat Detection Using FMCW Radar Based on Secondary Variational Mode Decomposition

This is the authors' abstract. We don't add key points for this paper.

In non-contact respiration and heartbeat detection using FMCW millimeter-wave radar, the effectiveness of signal separation directly determines the final detection accuracy. Traditional methods suffer from reliance on manual experience, inadequate suppression of high-order respiratory harmonics, and poor adaptability to complex environments. This paper proposes a signal separation algorithm based on Secondary Variational Mode Decomposition (SVMD). The Grey Wolf Algorithm (GWA) is used to adaptively select the key parameters of VMD for rough signal separation. After Wavelet Packet Decomposition (WPD), secondary VMD is applied to suppress high-order respiratory harmonics in the heartbeat signal, thereby achieving refined separation of respiration and heartbeat signals. Experimental results show that the proposed method achieves a detection accuracy of no less than 95.59% for both respiration and heartbeat in four scenarios, with a mean absolute error lower than that of traditional methods. The 95% Bland–Altman limits of agreement comply with the clinically referenced AAMI device accuracy expectations, demonstrating the high accuracy and strong stability of the proposed method.