Device-free Human Tracking Exploiting Phase Disturbances and Particle Filters

Abstract

This work presents a phase-based device-free localization (DFL) scheme for tracking the movement of a human target by employing a network of anchor RFID tags inside the search space. The proposed method measures the phase disturbances of the backscattered field induced by a target roaming in the propagation environment and detects whether an antenna-to-tag link is obscured or not. This binary information is fused by a geometrical model that realises the target as a cylinder, thanks to which the position of the target can be identified even by a single reader-antenna. A particle filter algorithm (PF) keeps track of the target's movement over time, smoothing otherwise abrupt maneuvers. Experimental results with a single antenna report a median absolute localization error in the order of 20cm which decreases to 15cm when more antennas are employed. The real-time capability of the method is also verified.