Projection Packet Collaborative Denoising: A Vital Sign Signal Reconstruction Solution for UWB Health Monitoring

Wireless wideband signals, such as ultra-wideband(UWB) and millimeter-wave signals, offer non-contact, high-resolution, and privacy-preserving detection, facilitating accurateand continuous vital signs monitoring. However, UWB signalssuffer from multipath and micro-motion noise, while heartbeatsignals are weak and susceptible to respiratory harmonics. Mostexisting methods model vital signs as single-frequency waves,ignoring multi-frequency characteristics. To address these chal-lenges, we propose a vital sign signal reconstruction frameworkbased on Projection Packet Collaborative Denoising (PPCD).By integrating wavelet packet decomposition with delay-domainmanifold projection denoising, PPCD effectively suppresses har-monic interference and micro-motion noise. It reconstructs sig-nals using multiple-frequency components rather than singlepeaks, achieving separation of overlapping spectra and precisetime-domain waveform recovery. R-peak detection then extractsrespiration rate (RR) and heart rate (HR). Evaluated in a week-long study with 10 subjects in office cubicles, we verified theadvantages of the proposed method in RR, HR, and heart ratevariability (HRV) analysis. Results demonstrate high accuracyfor RR (> 98%; average error: ±1 bpm) and HR (> 94%;average error: ±5 bpm) across varying settings. PPCD’s lowcomputational complexity enables efficient embedded implemen-tation, validated on a Lubancat4 board with 1.0–1.1 s processingper 60 s frame. By addressing spectral overlap and respirationharmonic interference, our approach improves RR/HR estimationaccuracy and robustness, supporting long-term, unobtrusive vitalsign monitoring in practical scenarios.