The CSI-ratio model is a signal processing method that divides the CSI measurements from two antennas or subcarriers by one another to cancel out shared noise components, hardware imperfections, and static multipath interference, thereby isolating the dynamic signal variations caused by human movement or presence. It matters for device-free sensing because it substantially improves the signal-to-noise ratio without requiring specialized hardware calibration, making passive human behavior recognition more robust and practical across diverse environments. Key variants include ratios computed across antenna pairs, subcarrier pairs, or combinations thereof, with some formulations operating in the complex domain to preserve phase information and others working on amplitude ratios to simplify computation.
Dictionary term