Design and Analysis of Realistic Mobility Models for Wireless Mesh Networks

Vehicular ad-hoc networks are a prospective technology which contribute to safer and more efficient roads. Inter-vehicle communication provides drivers and concerned authorities with information about road conditions, supports the coordination in case of emergency and improves traffic management. Furthermore, vehicular ad-hoc networks offer information and entertainment services to mobile users. Since large real-world testbeds are not feasible, research on vehicular ad-hoc networks depends mainly on simulations. Therefore, it is crucial that these simulations employ realistic mobility models. In this thesis, we design a generic mobility simulation framework (GMSF). The implementation of this framework greatly simplifies the generation of mobility traces and the evaluation of different mobility models. The modular design enables researchers to extend the framework with additional functionality. Furthermore, we propose a new vehicular mobility model representative of real-world characteristics. The GIS mobility model is based on highly detailed road maps from a geographic information system (GIS) and realistic microscopic behaviors (car-following and traffic lights). We perform an extensive comparison of our new GIS-based mobility model with popular mobility models (RandomWaypoint, Manhattan) and realistic vehicular traces from a traffic simulator (MMTS). In order to compare different mobility models by statistical means, we develop a set of mobility-related and graph-related metrics. Eventually, we investigate the influence of different mobility models on the performance of popular ad-hoc routing protocols (AODV and OLSR) in a network simulator.