An FMCW radar-based vital signs detection system using sparse Bayesian adaptive filtering achieved mean absolute errors of 0.53 Bpm for respiratory rate and 0.93 Bpm for heart rate.
A novel FMCW radar-based system using a sparse Bayesian adaptive filtering algorithm accurately estimates respiratory and heart rates in realistic environments.
Tasa de eventos absoluta: 0% vs 0%
Non-contact vital sign detection, particularly for respiratory rate (RR) and heart rate (HR), shows significant potential in healthcare applications due to its comfort advantages and privacy preservation. Nevertheless, detecting weak heartbeat signals remains challenging because of respiratory harmonics, ambient noise, and environmental clutter. To address this challenge, we propose a vital signs detection system based on frequency-modulated continuous wave (FMCW) radar. The system is built around a streamlined framework that combines four core modules designed to accurately extract RR and HR from the radar signals. The proposed system begins by removing static clutter using a Moving Target Detection (MTD) algorithm within the preprocessing module. Following this, the vital sign extraction module performs noncoherent integration to identify the distance bins most relevant to physiological activity. In the signal enhancement module, Pearson correlation analysis is used to combine signal regions with high consistency, while first-order phase differencing is applied to suppress baseline drift. This improves the stability and reliability of the extracted signals. To isolate respiratory and heart signals, the system employs a wavelet packet decomposition with autocorrelation (WPD-AC) algorithm. This approach effectively separates the respiratory and heart components while reducing noise and suppressing harmonic interference. Finally, a sparse Bayesian least-mean-squares (SBLMS) algorithm is introduced to dynamically refine prior information. This enables more accurate estimation of RR and HR. The experimental results show that the system achieves mean absolute errors (MAE) of 0.53 Bpm for RR and 0.93 Bpm for HR, demonstrating its notable capability in suppressing noise and respiratory harmonic interference. Furthermore, extensive experiments conducted in realistic office environments show that, compared with existing methods, the system exhibits excellent robustness across different users, distances, and angles conditions.
Wen et al. (Sat,) reported a other. An FMCW radar-based vital signs detection system using sparse Bayesian adaptive filtering achieved mean absolute errors of 0.53 Bpm for respiratory rate and 0.93 Bpm for heart rate.