Advanced Bayesian filtering techniques for UWB tracking systems in indoor environments

In this paper, we propose to use Bayesian filtering techniques so as to handle mobility and deal with biased times of arrivals (TOA) in indoor ultra wideband (UWB) positioning systems. More particularly, the modified extended Kalman filter (MEKF) and the modified regularized particle filter (MRPF) enable to track large biases affecting estimated radiolocation metrics due to LOS/NLOS and/or NLOS/NLOS transitions. Simulations were performed with a relevant deterministic UWB propagation model under realistic mobility assumptions. One important corollary for the appealing performances we obtained (e.g. the MEKF provides 90% of time with positioning accuracy better than 0.5 m in a "moderate walk" scenario) is that relaxed synchronization strategies could be adopted in UWB localizers. Indeed, the selection of secondary paths in the TOA estimation procedure, and hence, the introduction of large biases: 1) does not contradict the fine capability of UWB signals to resolve multipath profiles; 2) could enable to extend the range of UWB positioning systems in severe NLOS environments by alleviating the constraints on path detectability; and 3) does not degrade significantly the positioning accuracy if the spatial correlation of TOA biases is taken into account.