Two-channel VMD Method for Detection of Respiratory and Heart Rates with MIMO MMW Radar

Non-contact intelligent sensing of humans based on millimeter wave (MMW) radar is of significant value in the field like automatic pilot and intelligent medicine. This paper presents a method of two-channel variational mode decomposition (VMD) to implement the detection of the respiratory rate (RR) and heart rate (HR) of humans for multiple-input multiple-out (MIMO) MMW radar. Specifically, the two channels are selected by peak searching in the frequency domain and retain high energy in the selected channels with peak frequencies in the normal range of RR or HR. For the VMD operation, the number of decomposition layer is set to three to better extract components of respiration, heartbeat and the beat signal (or its high-order harmonics) from the signal after vital signs filtering (VSF). In regard to the intrinsic mode functions (IMF) after two-channel VMD processing, correlation processing is used to further enhance the accuracy of detection, which retains the IMF of same frequency as the RR or HR for subjects. Finally, the experimental results validate that compared with the existing time-frequency domain estimation, two-channel correlation and single-channel VMD, this method improves the detection precision and stability of RR and HR when the radar is arranged in frontal and rear direction of human.