Particle Filters for Positioning WiFi Device Users

摘要

In this paper, we present an effective particle filter-based approach to positioning WiFi device users in indoor environments. In order to track a user in a continuous space, we assume that the position of the user corresponds to a specific point on an edge of the graph, and suggest a graph-based representation of the continuous state space. Enabling to compute the measurement likelihood of the arbitrary points, including the points where any training data has never been collected before, we present a linear interpolation-based perception model. Unlike mobile robots, it is difficult to obtain the exact motion control data from pedestrians. We suggest a non-linear state transition model for representing the motion of a WiFi device user, which consists of three different component models. We explain the implementation and experiments for investigating the performance of our particle filter-based localization method.