A Study of Environment Noise in Ultra-Wideband Indoor Position Tracking

This work is motivated by the problem of improving the accuracy of indoor ultra-wideband (UWB) position tracking through the study of the environment noise that affects such a system. Current systems can provide accuracy in the range of 30-100 cm in a small building, suitable for applications that require rough room-level precision such as asset tracking and surveillance. Our long-term goal is to improve the accuracy to 1 cm or better, expanding potential applications to telepresence, augmented reality, training and entertainment. This work investigates the possibility of systematically observing the measurement noise of an UWB position tracking system and building a map of it throughout a facility. In order to understand the effect of environment noise on UWB indoor positioning and in turn filter out the effects of this noise, it is important to have an idea of what this measurement noise looks like in a real world scenario. In this work, an understanding of the measurement noise is gained by taking many measurements using a commercially-available UWB positioning system installed in a real world scenario and analyzing these measurements in various ways. To the author’s knowledge, no one has used such an exhaustive approach to analyze measurement noise in UWB indoor positioning. The results of this work show that the measurement noise that affects a UWB indoor position tracking system can be effectively modeled using a weighted sum of Gaussians, is stable over time and is locally similar. Furthermore, a particle filter augmented with a measurement noise map is proposed to improve position tracking accuracy. Finally, a metric is proposed that can be used to quantify expected system performance based on sensor location, sensor orientation and facility floorplan. Using this metric, a procedure is developed to determine the parameters, i.e. sensor position, sensor orientation and potentially others, of the physical installation of the UWB tracking system that will produce minimum measurement error based on sensor geometry and physical facility constraints.