A Novel Approach to Accurate Respiratory Rate and Heart Rate Estimation via FMCW Radar

Abstract

Vital signs, such as respiratory rate (RR) and heart rate (HR), are essential indicators for assessing human health. Radar enables noncontact detection of RR and HR. However, chest displacement caused by heartbeats is much smaller than that caused by respiration. And the weak heartbeat signal is susceptible to being overwhelmed by respiratory harmonics and noise, making HR detection challenging. To address these issues, we propose a novel vital sign decomposition method. Phase difference and Hampel filter are used to suppress unknown noise, and discrete wavelet transform (DWT) is used to separate respiratory signal. To achieve more accurate estimation of heartbeat signal frequency, we introduce the singular value decomposition (SVD)-focal underdetermined system solver (FOCUSS)-recovery (SFR) method following successive variational mode decomposition (SVMD). This method possesses feature extraction and sparsity optimization capabilities, thereby improving spectral resolution and the estimation of heartbeat signal frequency. Experimental results demonstrate that the root mean square error (RMSE) between ECG sensor measurements and proposed method is below 1 beat per minute (bpm) for RR and below 2 bpm for HR.