A multi-scale Signal Quality Assessment framework for RF-based vital sign sensing accurately reflected underlying physiological signal quality and correlated strongly with downstream estimation accuracy.
A novel multi-scale Signal Quality Assessment framework improves the reliability of contactless RF-based physiological sensing by accurately quantifying static and dynamic interference.
Contactless physiological sensing using radio frequency (RF) signals has shown great promise in enabling unobtrusive health monitoring. However, assessing the reliability of RF-derived physiological measurements without ground-truth references remains underexplored. The low-quality RF signal containing environmental interferences may be mistakenly interpreted as valid physiological activity, leading to inaccurate health indicators or misdiagnoses. To address the challenge of highly entangled interference and physiological variations, existing studies rely on heuristic rules or motion detection to filter poor-quality data, lacking an interpretable and generalizable framework for evaluating RF signal quality. This paper proposes a multi-scale Signal Quality Assessment (SQA) framework for RF-based vital sign sensing. Our approach revisits the fundamentals of wireless sensing and performs short-term IQ-domain analysis to separate physiological motion from interference. By applying circle fitting and respiration reconstruction across multiple temporal scales, the proposed framework independently quantifies static and dynamic interference for both respiration and heartbeat signals. These measures are then integrated into a unified quality score, providing a robust and interpretable indicator of signal reliability. Experiments on RF recordings from over 4,500 individuals and on 10 full-night recordings demonstrate that the proposed framework accurately reflects the underlying physiological quality of the signals. The resulting quality scores show strong correlations with downstream estimation accuracy of respiration motion, heart rate, inter-beat interval, and heart rate variability.
Gong et al. (Mon,) conducted a other in Physiological sensing (n=4,500). Multi-scale Signal Quality Assessment (SQA) framework was evaluated on Estimation accuracy of respiration motion, heart rate, inter-beat interval, and heart rate variability. A multi-scale Signal Quality Assessment framework for RF-based vital sign sensing accurately reflected underlying physiological signal quality and correlated strongly with downstream estimation accuracy.