UWB Radar for Health Monitoring: A Multi-Frequency Method for Vital Sign Signal Reconstruction

Cardiovascular risk is increased in modern offices  or studies due to stress, irregular schedules, lack of exercise,  and persistent cardiac arrest in high-intensity environments. Traditional contact-based methods offer high accuracy but suffer  from poor long-term usability (frequent charging, wearing discomfort,  life-unfriendly). Ultra Wideband (UWB) signals offer  non-contact detection, high resolution, strong penetration, and  privacy preservation, which facilitates accurate and continuous  vital signs monitoring. However, UWB signals are affected by  multipath\/micro-motion noise, and the heartbeat signal is weak  with susceptible to respiratory harmonics. Most existing methods  simplify vital signs as single-frequency sinusoidal waves, ignoring  their essential multi-frequency characteristics and failing to  separate the respiration and heartbeat when the respiration harmonics  overlap with the heartbeat band. We propose a vital sign  signal reconstruction framework based on Projection-Packet Coordinated Denoising (PPCD). This framework combines wavelet  packet decomposition and signal subspace projection denoising  techniques in the delay domain to mitigate harmonic interference. It can reconstruct the respiration and heartbeat signals in the  frequency domain using multiple frequency components instead  of relying on a single frequency peak. Thus, the separation  of overlapping spectra and the precise reconstruction of timedomain  waveforms of multi-frequency signals can be achieved. R-peak detection and RR-interval analysis are further applied  to extract respiratory rate (RR)\/heart rate (HR), and analyze  heart rate variability (HRV) for physiological assessment. Evaluated  in a week-long study with 10 subjects in office cubicles (over 2400 minutes of data), we verified the advantages of the  proposed method in RR, HR, and HRV analysis by comparing  two baseline methods. It demonstrated the reliable performance  under different distances and posture changes. By effectively  suppressing respiratory harmonics, our approach improves respiratory\/  heartbeat estimation and enables long-term, unobtrusive  vital monitoring under complex scenarios.