Year: 2025
Type: article
Abstract: Recent years have witnessed significant advancements in contact-free vital signs monitoring, yet how to clearly separate distinct vital signs (e.g., respiration and heartbeat) co-located on a human body remains largely open. Prior art typically leverages complex post-processing techniques, such as sophisticated filters and even deep learning models, but their performances are often questionable given their non-robust outcome and high computational complexity. To this end, we propose to exploit fine-grained physical resolution to act as a pre-processing threshold, so that a targeted vital signs waveform can be distilled out of interferences from both body motion and other vital signs; we then construct Sonic-Fi as an ultrasonic sensing platform to realize this ambition. Sonic-Fi is built on innovatively crafted hardware to achieve a tunable resolution at mm-level, which serves as a physical threshold to separate two co-located waveforms with distinct amplitudes. This separation further enables us to design algorithms that recover the beyond-threshold waveform with amplitude interpolation while obtaining the below-threshold one via phase tracing. The same concept is finally adapted to eliminate the effects of body movements. Extensive evaluations demonstrate that Sonic-Fi is capable of simultaneously recovering respiratory and heartbeat waveforms accurately, significantly promoting contact-free vital signs monitoring for real-world adoption.
DOI: 10.1145/3749460
Authors: Penghao Wang, Shujie Zhang, Chao Cai, Chao Liu, Jun Luo
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Award: G2738460470
OpenAlex ID: W4413968588