Research on multi-person respiration and heartbeat detection method based on adaptive VMD using mmWave radar

To address the challenge of distinguishing individual human targets and separating their respective respiration and heartbeat signals in complex environments using millimeter-wave radar, this paper proposes a multi-person respiration and heartbeat detection method based on adaptive variational mode decomposition. First, spatial smoothing technology is introduced to improve the MUSIC algorithm, enhancing the angle resolution of the millimeter-wave radar and enabling accurate localization and differentiation of multiple targets. After determining the positions of each target, the corresponding phase signals are extracted, followed by phase unwrapping and phase differencing. Finally, considering the spectral characteristics of human respiration and heartbeat signals, an adaptive variational mode decomposition method is implemented by setting a threshold for the energy ratio of the frequency band. This approach avoids the limitation of the traditional VMD method, which relies on subjective experience to select the number of decomposition modes, and extracts the corresponding respiration and heartbeat signals from the separated IMF components. Experimental results show that the proposed method effectively detects respiration and heartbeat signals of each target in both dual-person and triple-person scenarios, with a frequency error range superior to traditional methods, achieving high detection accuracy.