Localization and Mobility Modeling in Wireless Ad Hoc Networks

Abstract

The availability of wireless devices and their increasing ability of sharing information has stimulated the emergence of the wireless ad hoc networks. In this dissertation, we focus on two topics related to wireless ad hoc networks. One is the localization in the static wireless sensor networks and the other one is the mobility modeling in the mobile ad hoc networks.

Localization is a fundamental service for many applications of wireless sensor networks. In the first part, we consider a distributed, probabilistic localization approach, which is suitable for systems with inaccurate range measurements and a small number of beacons. Two solutions are provided: an RSS-based solution and an angle-based approach corresponding to the measurements used between neighbor nodes. The basic idea of these approaches is to restrict the possible locations of the nodes by using probabilistic constraints. The proposed probabilistic approach is evaluated through simulations based on real-world and simulated measurements; the results are compared with the Cramer-Rao lower bound and other RSS-based and angle-based localization algorithms. The results show that, with inaccurate range measurements, and a small number of beacons, the proposed probabilistic approach outperforms existing methods and approaches the optimum bound.

Due to the scarcity of real mobile ad hoc deployments, most protocol evaluations are carried out through simulations. One of the core components or network simulations is the mobility model, which characterizes the mobility patterns of the mobile devices in a network and uses such patterns to reproduce trajectories of the mobile devices accurately. In the second part of this dissertation, we study several real world traces including wireless LAN and bus traces. The statistical properties of several features that primarily determine the behavior of the mobile users are extracted and thoroughly analyzed. Using the extracted statistical properties and the statistical similarities of those features, we propose a simple generalized mobility generator that can generate both realistic and diversified synthetic traces.