Sequential Monte Carlo for simultaneous passive device-free tracking and sensor localization using received signal strength measurements

This paper presents and evaluates a method for simultaneously tracking a target while localizing the sensor nodes of a passive device-free tracking system. The system uses received signal strength (RSS) measurements taken on the links connecting many nodes in a wireless sensor network, with nodes deployed such that the links overlap across the region. A target moving through the region attenuates links intersecting or nearby its path. At the same time, RSS measurements provide information about the relative locations of sensor nodes. We utilize the Sequential Monte Carlo (particle filtering) framework for tracking, and we use an online EM algorithm to simultaneously estimate static parameters (including the sensor locations, as well as model parameters including noise variance and attenuation strength of the target). Simultaneous tracking, online calibration and parameter estimation enable rapid deployment of a RSS-based device free localization system, e.g., in emergency response scenarios. Simulation results and experiments with a wireless sensor network testbed illustrate that the proposed tracking method performs well in a variety of settings.