Rethinking RSSI for WiFi sensing

Le résumé fourni par la source

The Received Signal Strength Indicator (RSSI) is available on WiFi devices, yet it is often regarded as too coarse for sensing. This work revisits its sensing potential and presents WiRSSI, a bistatic framework for RSSI-only passive tracking and motion sensing. Moving beyond path-loss RSSI models, we develop a Channel State Information (CSI)-power-based formulation to model RSSI as an energy-domain observable, and show that it preserves motion-dependent signatures enabling Doppler, Angle-of-Arrival (AoA), and delay cue extraction. WiRSSI uses one transmitter and one receiver with a three-antenna array, exploiting antenna diversity to suppress the mirror symmetry of real-valued RSSI. WiRSSI estimates Doppler and AoA via low-complexity joint time and space spectral analysis and infers bistatic delay from amplitude-only information. The AoA and delay estimates are mapped to Cartesian coordinates and regularized to recover trajectories. Experiments on commodity network interface cards achieve median and 90th-percentile XY errors of 0.825 m and 1.509 m, respectively, versus 0.562 m and 1.047 m from a typical CSI-based method. We also demonstrate RSSI-only gesture recognition, showing that WiRSSI features retain discriminative motion cues. Overall, RSSI can support practical sensing as a complementary option when CSI access is restricted or when fine-grained CSI is privacy-sensitive.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.