PUB - Publikationen an der Universität Bielefeld

This dissertation addresses a bidirectional scheme of Ultra-Wideband (UWB)-based localization system, which is generally overlooked in the literature causing its potential overshadowed. The bidirectional scheme is one of the three design integration processes usually applied in UWB-based localization systems. The key property of the bidirectional UWB scheme is its ability to act as both navigation and tracking tasks within a single localization scheme. Conventionally, the perspective of navigation and tracking in wireless localization systems is treated separately, because distinct methodologies were typically required in the implementation process. The ability to combine two unique positioning perspectives (i.e., navigation and tracking) within a single scheme is, indeed, a paradigm shift in the way location-based services are observed in the literature. Much to the author's surprise, there are no well-documented books or research articles related to the bidirectional UWB localization system. Thus, this dissertation attempts to serve as a complement to the mentioned gap in the literature. In this dissertation, the bidirectional UWB localization scheme was tackled by dividing the system implementation process into several sectors. Then, the methodologies applicable in each sector were rigorously evaluated in order to give the readers insightful knowledge, findings, and recommendations. Regarding this, the concept of the bidirectional~UWB localization scheme and its implementation process were thoroughly explained by comparing it with the typical unidirectional schemes. Concerning the ranging sector of the bidirectional UWB scheme, this dissertation demonstrated the misconception widely practiced in literature in terms of the two-way ranging technique. Moreover, the dissertation suggested a better method compared to the conventional one, which could be used as a baseline or de facto standard for comparing or bench-marking different two-way ranging schemes. The claim was supported by the experimental results rigorously evaluated by using analytical methods, numerical simulation, and real-world experimental data. Moreover, the comprehensive benchmark of five location estimation algorithms for the bidirectional UWB localization system was conducted in the dissertation. The five algorithms were evaluated based on the Bayesian framework and the detailed implementation process was regarded as an important aspect because the literature is generally lacking it, especially for the use-case of the UWB technology. The evaluation results showed that the linear positioning algorithms gave excellent performance in scenarios such as static conditions under a direct path signal. In contrast, the nonlinear techniques appeared to be better at resisting abrupt changes during measurements as well as the scenarios in non-direct path signals. Furthermore, the identification and mitigation of errors produced by non-direct path signals in UWB were addressed. Concerning this, a novel mitigation technique was proposed in the thesis whereas the feasibility of the identification process was evaluated using machine learning methods. The classification procedure was considered as a multi-class problem, which is opposed to the typical binary class approach in literature. The results showed that the machine learning techniques are very promising compared to the conventional ones for identification of the non-direct path signals in UWB localization.