Radar-Based Vital Signal Extraction via Morphological Component Analysis

• Published in: IEEE transactions on microwave theory and techniques pp. 1 - 16
• Main Authors: Li, Hongzhi, Ayuknso, Ayukocha Gandhi Bessemntoh, Yang, Nanyi, Wei, Zeying, Zhi, Minghan, Hua, Qinglong, Zhao, Bin
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: Accuracy; Biomedical monitoring; Clutter; Frequency estimation; Frequency-modulated continuous wave (FMCW); Monitoring; morphological component analysis (MCA); Motion artifacts; Radar; Radar tracking; remote health monitoring; Tracking; Transforms; vital signal detection
• ISSN: 0018-9480; 1557-9670
• DOI: 10.1109/TMTT.2025.3590047

Noncontact monitoring of cardiopulmonary activity using radar has been a long-standing challenge, especially in dynamic environments where motion artifacts and environmental clutter significantly degrade signal quality. This article introduces a novel signal processing framework for accurate estimation of respiration rate (RR) and heart rate (HR) using frequency-modulated continuous-wave radar. The proposed approach integrates adaptive beamforming for spatial filtering, dual constant-Q rational dilation wavelet transform (RADWT) for morphological component analysis (MCA), and synchrosqueezed time-frequency ridge tracking to extract and track vital sign information. By leveraging the distinct morphological characteristics of physiological signals and motion artifacts, the dual-RADWT-based MCA effectively separates the oscillatory components associated with RR and HR from transient disturbances, even in the presence of spectral overlap. The refined time-frequency representations (TFRs), enhanced through ridge tracking, ensure stable and accurate tracking of the vital signals. Comprehensive evaluations across multiple subjects and scenarios demonstrate that the proposed method consistently outperforms conventional techniques. The results show that the proposed framework achieves significantly lower mean absolute errors (MAEs) and root-mean-squared errors, highlighting its robustness and precision in real-world health monitoring applications, even in clutter-rich environments.