Geographische Restriktionen bei menschlicher Bewegungsmodellierung zur Leistungsbewertung in mobilen Netzen


Before deployment in a productive system, simulation is the most prominent choice to efficiently evaluate the performance of mobile communication networks. Simulation of these networks requires models which reflect relevant parts of a real deployment for computer processing. Mobility models reflect mobility patterns and are known to have a significant impact on the performance evaluation of mobile networks. However, in the literature, modeling of human mobility is characterized by so-called flights, i.e., direct lines from start to end point. Therefore, the consideration of geographic restrictions in the context of human mobility modeling has been mostly neglected so far. In this dissertation thesis, we consider geographic restrictions during multiple phases of trace-based modeling of human mobility while focusing on simulative performance evaluation of mobile communication networks. First, we identify and solve several challenges arising during the general modeling design of these restrictions. The map-based approach chosen for realizing geographic restrictions in combination with these solutions has led to the implementation of a general framework for map-based mobility models. In order to evaluate the impact of the map-based approach on the simulative performance evaluation in mobile networks, we further extend the state-of-the-art mobility model SLAW by geographic restrictions to MSLAW and compare the flight-based original to the map-based extension in the context of various performance metrics. We then show as the main result of the evaluation part of this work that geographic restrictions indeed significantly impact contact-based metrics. Motivated by this result, we finally inspect the utilization of geographic restrictions during an earlier modeling phase. The proposed map-oriented approach to trace processing alleviates several problems of the GPS trace data under consideration. Thus, considering geographic restrictions already during trace processing as preparation for trace analysis is also a sensible choice.

Dissertation Thesis (in German), University of Osnabrück, GCA-Verlag, ISBN 978-3-89863-267-6.