MultiSense: Enabling Multi-person Respiration Sensing with Commodity WiFi
summary
MultiSense uses CSI from a single pair of commodity WiFi transceivers (three antennas each), models received signals as linear mixtures, and separates individual breaths with independent component analysis (ICA) after removing amplitude impulse noise and time-varying phase offsets by creating a weighted-antenna nulling denominator that preserves linearity. Prototype extracts continuous respiration waveforms for up to four people, resolves similar rates and blind spots, and reports mean absolute respiration rate error 0.73 bpm.
AI-written summary · about