UWB for Robust Indoor Tracking: Weighting of Multipath Components for Efficient Estimation

In a radio propagation channel, deterministic reflections carry important position-related information. With the help of prior knowledge such as a floor plan, this information can be exploited for indoor localization. This letter presents the improvement of a multipath-assisted tracking approach using information about the relevance of deterministic multipath components in an environment. This information is fed to a tracking filter as an observation noise model. It is estimated from a few training signals between anchors and an agent at known positions. Tracking results are presented for measurements in a partial non-line-of-sight environment. At a bandwidth of 2 GHz, an accuracy of 4 cm can be achieved for over 90% of the positions if additional channel information is available. Otherwise, this accuracy is only possible for about 45% of the positions. The covariance of the estimation matches closely to the corresponding Cramèr-Rao lower bound.