Stochastic Data Association for Multipath Assisted Positioning Using a Single Transmitter

Abstract

This paper builds on and extends the Channel-SLAM algorithm which exploits a multipath radio channel for the position estimation of mobile receivers. Channel-SLAM treats Multipath Components (MPCs) as Line-of-sight (LoS) signals originating from Virtual Transmitters (VTs) and estimates the positions of VTs and receiver simultaneously based on Bayesian filtering. The current Channel-SLAM implementation does not involve the retracking of previous MPCs or VTs. Therefore, when the tracking of an MPC is lost and, subsequently, regained, the corresponding VT is initialized without any prior information. Incorporating a stochastic data association algorithm extends Channel-SLAM and enables the retracking of VTs even when the MPC has been lost. The proposed algorithm increases positioning reliability, decreases computation complexity, and improves the precision of Channel-SLAM. Additionally, this paper presents a novel transition model using inertial sensors for a hand-held device for moving pedestrians. The developed positioning algorithm is evaluated based on measurement data obtained in an indoor scenario using an off-the-shelf Ultra-WideBand (UWB) module. Evaluations show that accurate position estimation can be done using only one physical transmitter and without requiring any knowledge of the physical transmitter position.