6 November 2023 in IEEE Journal of Electromagnetics, RF and Microwaves in Medicine and Biology
Vital signs are objective indices of health status. Heart rate variability (HRV) is the physiological phenomenon of variation in interval between consecutive heartbeats, which carries more physiological information than respiration rate (RR) or heart rate (HR). Medical radar isa potential sensor for non-contact vital sign monitoring. However, this sensor requires a more complicated process for HRV extraction. In this study, the HRV was obtained as follows: The heartbeat component was extracted from the radar signal using a locally projected noise reduction (LPNR)-based nonlinear adaptive filter and a convolution-based smoothing filter. The heartbeat component still includes undesired peaks; therefore, a derivative function was used to detect the sharpest slope peak, which is the desired peak, followed by R-peak detection to obtain HRV. To evaluate the performance of the proposed method, we tested the system on 18 healthy subjects and compared the HRV determined by this sensor with that measured by contact-type electrocardiography (ECG). The results show a correlation of 97.43% between HRV by radar and HRV by ECG; the 95% confidence of inter-beat interval (IBI) is 59.5 ms. In addition, the proposed method was applied to monitor the change in HRV of an inpatient from Yokohama Hospital, Japan. The clinical data processing results provided consent for the nurse's daily check.
30 July 2025 in IEEE Access
Radar technology, particularly frequency-modulated continuous wave (FMCW) radar, has garnered attention in the field of smart home monitoring owing to its high sensitivity, long-range surveillance capabilities, and privacy-preserving characteristics. This study proposed a human motion state recognition system to recognize motion states based on the spectra from radar signals in a home environment. The velocity of a target is assessed by comparing a micro-Doppler signature against a velocity threshold, and its height signature discerns between high and low postures against a height threshold. The horizontal position of the target is determined using a range–angle map. The proposed threshold algorithm defines the motion state using the velocity, height postures and indoor position. Subsequently, the performance of the system is evaluated by conducting experiments on 10 subjects. The results demonstrate the efficacy of the proposed method, achieving an accuracy of approximately 85%. In continuous-action experiments involving sequences of movements, the system achieved recognition accuracy exceeding 90%.
1 November 2026 in IEEE Sensors Letters