Side effects in ultrawideband indoor positioning systemspresence of the human body

  1. Otim, Thimothy
Supervised by:
  1. Alfonso Bahillo Director
  2. Francisco Falcone Director

Defence university: Universidad de Deusto

Fecha de defensa: 16 December 2020

  1. Domenico Giustiniano Chair
  2. Álvaro Hernández Alonso Secretary
  3. Estefanía Muñoz Díaz Ropero Committee member

Type: Thesis

Teseo: 660074 DIALNET


Global Navigation Satellite Systems (GNSS) are the main sources of location information, but mainly for outdoor environments. However, several Indoor Positioning Systems (IPS) relaying on alternative sources of information have been developed because the satellites of GNSS require line-of-sight to operate continuously, which cannot be guaranteed indoors. Key among the alternative technologies for IPS is Ultrawideband (UWB), thanks to its unique characteristics such as immunity to fading, multipath resolving capability, short time pulses and accurate detection of the first pulse, among others, which make it attractive for time of flight (TOF) ranging and subsequent positioning. However, despite the potential UWB seems to offer, it can be a challenge to achieve accurate positioning in pedestrian tracking contexts because of the Non-line-of-sight (NLOS) situations created by the body of the user carrying or wearing the sensor. The blockage of the direct path between the moving and fixed sensors (known as anchors) created by the pedestrian’s body is known as human body shadowing and can lead to position errors of up to several meters. Currently, there exist limited proposals in the literature that address the sides effects in UWB IPS provoked by the presence of the human body in a specific and comprehensive manner, yet a solution to the human body shadowing problem would make UWB more attractive for pedestrian localization. Therefore, the main goal of this thesis is to analyse and mitigate the NLOS provoked by the human shadowing in stand-alone UWB IPS to keep the accuracy below one meter even in NLOS situations, which is the reasonable accuracy requirement for pedestrian tracking applications. To this end, five topics have been investigated: i) UWB radiation and human body interaction using the finite difference time domain (FDTD) technique, ii) analysis of the human shadowing effect on the received signal strength (RSS) and TOF ranging and subsequent positioning accuracy, iii) a proposition of UWB TOF ranging error models, iv) evaluation of an algorithm that mitigates the human shadowing effect using the previous error models, and v) modelling of a time-frequency channel approach for UWB ranging applications using a 3D Ray launching simulation technique. In relation to the effect human shadowing imposes on the ranging and subsequent positioning accuracy, up to seven wearable sensor positions have been analysed through an extensive measurement campaign. Two research questions are answered: Which is the best and worst positions on the human body where to mount a wearable sensor? and which are the wearable positions that generate the positioning accuracy greater than the required one meter? In addition, the UWB ranging error model for each wearable sensor position is built by exploiting the relative position among the pedestrian, anchor, and wearable sensor. By combining the ranging error models into a particle filter, an algorithm that can mitigate NLOS effects for different wearable sensor positions that generate positioning accuracy above one meter is devised. This algorithm reduces the position error by up to 75 % and 82 %, as shown by the results obtained through simulations and real experiments, respectively, leading to the achievement of a sub-meter level of localization accuracy. If multipath propagation is present, channel estimation becomes important for time delay-based positioning methods. Considering multipath conditions in an indoor environment with complex scatterer distributions, a valuable tool to test UWB systems for ranging applications with a mean accuracy of up to 10 cm has been developed. In addition, a parametric study has been performed considering variations of cuboidsize resolution of the simulation mesh, to analyse the convergence impact on estimation accuracy, focusing on radio frequency power levels as well as time domain characterization. The results obtained throughout this thesis indicate that clear weaknesses caused by NLOS effects due to the human body exist, and still need to be addressed. More simulations and measurements in different environments can be performed to complement and generalize the main conclusions which have been drawn. Therefore, work must continue to improve the pedestrian UWB IPS, for instance through online signal processing of the error models which can be utilized in mitigating the human body shadowing effect. This may be key for integrating the solution in a mass market device.