Robust UWB Indoor Position Tracking Using TDOA Measurements

Abstract

This paper considers the problem of ultra-wideband (UWB) indoor position tracking using time difference of arrival (TDOA) measurements. The target broadcasts UWB pulses and their time of arrivals (TOAs) are estimated at spatially distributed base stations to extract the received TDOAs for position tracking. In indoor environments, the measured TOAs may be subject to errors due to the multi-path and/or non-line-of-sight (NLOS) propagation of the UWB pulses. This could lead to large TDOA noises and severely degraded tracking performance. To address the aforementioned challenge, a new two-stage indoor tracking algorithm is developed. The first stage preprocesses the sequentially obtained TDOA measurements using an outlier-robust Kalman filter. The impact of the large TDOA errors is mitigated by assuming that the TDOA noises follow the heavy-tailed Student's $\boldsymbol{t}$ -distribution. The filtered TDOAs are then fed into the second stage that utilizes a grid-based filter to estimate the current target position. The state space covered by the grid-based filter is a rectangular region that is updated recursively via bound expansion and contraction operations. The use of the recursive bound further decreases the degrading effect of NLOS TDOA noises on the position tracking accuracy. Experiments using synthetic and real-world data illustrate the good performance of the newly proposed UWB indoor tracking algorithm.