This Is AuburnElectronic Theses and Dissertations

Maximizing Connectivity and Performance in Mobile Ad Hoc Networks Using Mobile Agents

Date

2007-12-15

Author

Dengiz, Orhan

Type of Degree

Dissertation

Department

Industrial and Systems Engineering

Abstract

Mobile wireless ad hoc networks are instantaneous, autonomous telecommunication networks that provide service to users wherever and whenever the service is needed. The communication depends on wireless links that are formed between the users. A link is formed between two users if they are within each other’s wireless communication range. The mobility in these networks can cause links to disconnect, disrupting communications. A new strategy is proposed which controls the movements of some mobile agents to maintain network connectivity. The main objective of these mobile agents is to maximize network data flow, which is formulated as an all-pair maximum flow problem. This is accomplished by optimizing the movements of the agents to their next locations as the user nodes travel freely in the field. The representation of ad hoc network performance in terms of an all-pair maximum flow problem is novel as is dynamically optimizing the agent nodes using heuristic algorithms integrated with network flow algorithms. Two evolutionary inspired, population based heuristic algorithms; a genetic algorithm and a particle swarm are developed along with an approximate linear programming model as optimizer tools. The results show the advantage of employing heuristic algorithms due to the complexity of the problem. While the approximate linear model could only solve small static and medium dynamic problems with poor results, the heuristics performed successfully for problems two to four times larger. These heuristic approaches will enable robust and physically self organizing networks with superior connectivity properties. The approach proposed in this research can be applied to static scenarios and dynamic situations. This is important because there are practical static applications of ad hoc networks, mainly in sensor networks. The novel models and algorithms developed should enable new research and and commercial opportunities in ad hoc wireless networking.