Authors: Thorsten Wehs, Allan C. Maheri, Carsten Koch, Tilman Leune
In: accepted atThe 12th Workshop on Positioning, Navigation and Communication 2015 (WPNC'15) in Dresden, Germany, March 11-12, 2015
Abstract: One of the applications of Ultra-wideband pulses are in the area of indoor localization systems whereas the anchor nodes are placed at stationary positions on the ceiling or walls, while the mobile nodes can be located at any given location inside the building. The localization system has to estimate the location of the mobile nodes, but when the anchor nodes and the mobile nodes are separated by a solid material, such as a wall, wood, partition and glass, the performance of a localization system deteriorate due to distortion of UWB pulses as it penetrates the material. The quality of distortion is mainly affected by the type of material, thickness and the pulses’ angle of incidence.
In this paper, we analyse features which strongly characterizes the distorted pulses and with help of machine learning techniques, we present a framework for estimating the material type, material thickness and pulse’s angle of incidence from a received distorted pulse. First, the material which distort the UWB pulse is classified, second the material thickness is assigned to four given classes of thickness and finally the pulse’s angle of incidence is estimated.