Modeling and Simulation of Ultra-Wideband Indoor Localization Systems in Soft-Non-Line-of-Sight

Within the last decade Ultra-Wideband (UWB) has gained a lot of interest. Especially, localization systems are exploiting the high bandwidth and thus high time resolution of impulse based UWB systems. The applications for UWB systems are diversified, from high-data rate personal area networks over sensor networks to emergency communication with localization capabilities. For indoor localization the main research focus was on time of arrival estimation and trilateration methods, which motivated this thesis to answer, if UWB localization can benefit from DoA estimation in indoor scenarios. In indoor scenarios two main issues complicate the localization. The transmitted signals reach the receiver over a huge number of multipath. Additionally, the probability of non-line-of-sight between transmitter and receiver is high in indoor scenarios which has to be accounted for.

Today, the huge computing power of CPUs enables to simulate even computationally complex mobile communication and localization systems to test new algorithms and predict their behavior without the initial need for expensive hardware and measurement campaigns. Here, a crucial point is the accuracy of the deployed models. The results and predictability of the algorithms can only be as good as the applied models. It is noticed, that often the used channel models, especially for localization system simulation, can be improved. The used channel models are mostly based on stochastic processes, but for localization system simulation site specific channel models should be the preferred choice. On the other hand MIMO system simulation require sophisticated channel models as well in order to allow for reliable results.

UWB channel modeling is an area with a lot of research activities showing the demanding task of accurately predicting the UWB channel characteristics. First started with investigations of UWB localization and MIMO algorithms, this pointed to the requirements for channel simulation and lead to research activities in ray tracing techniques with focus on frequency selective indoor scenarios including precise modeling of transmission effects.

Based upon ray tracing several investigations of channel parameters like time and direction of arrival of the direct path in soft-non-line-of-sight scenarios and their influence on the localization accuracy were performed. The results underline the potential of not only using time of arrival estimates but also direction of arrival estimates to improve UWB indoor localization. This motivated the introduction of two algorithms for direction of arrival estimation. The algorithms are analyzed in a system level simulation including channel impulse responses simulated with the indoor ray tracer.

The thesis is structured as follows. Chapter 2 gives an overview about UWB, channel impulse response, MIMO systems and localization. The developed ray tracer is introduced in chapter 3 and channel measurement are evaluated. UWB beamforming and the comparison to narrowband beamforming is investigated in chapter 4. In chapter 5 the ray tracer is applied to analyze errors on localization caused by signals transmitting through a wall. Next, different parameter estimators are introduced in chapter 6. Here, a novel algorithm, BeamLoc, is presented to estimate the direction and time of arrival of a signal. Simulation results for three different scenarios and three different system configurations are shown in chapter 7. All simulations are done with Matlab. A conclusion and outlook is given in chapter 8.