Akademik Veri Yönetim Sistemi
IEEE SENSORS JOURNAL, cilt.25, sa.23, ss.43208-43222, 2025 (SCI-Expanded, Scopus)
Due to its multiobjective potential for noncontact vital signs' monitoring, millimeter-wave (mmW) radar has increasingly drawn attention in human health and safety-related sensing applications. However, detection of vital signs, especially weak heartbeat reaction, is more challenging when disrupted by interference from background noise, random human body movement, and the sensitive nature of radio waves. To address these problems, this work presents an improved frequency-modulated continuouswave (FMCW) radar vital signs' monitoring solution incorporating phase error correction and heartbeat event probability prediction. The main contributions include the following: 1) development of a data processing framework reinforcing radar echoes for high signal-to-noise ratio (SNR) vital signs' detection, which amplified the returned signals through beamforming and compensated phase perturbation; in addition, two techniques including adaptive mode decomposition and neural network have been cordially adopted to perform signal conditioning; 2) proposal of a phase error correction method with an adaptive dual-sliding window to mitigate the phase noise and distortion introduced by the nonperiodic body movement, nonstationary breathing pattern, dynamic environmental clutter, and so on; it overcomes susceptibility to noise for the phase response and improves its stability and continuity; and 3) establishment of a deep neural network (DNN)-based model to predict the probability distribution of heartbeat events with phase segmentation. This prediction model avoids rigid misclassification of heartbeats and enhances the algorithm's tolerance to noise and adaptability to complex conditions. Experimental results have verified the effectiveness of the proposed solutions. The presented method provides a robust solution for reliable, high-accuracy, and continuous vital signs' monitoring in real environments.