Date of Award

Spring 2004

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Electrical Engineering - (Ph.D.)

Department

Electrical and Computer Engineering

First Advisor

MengChu Zhou

Second Advisor

Constantine N. Manikopoulos

Third Advisor

Hongya Ge

Fourth Advisor

Symeon Papavassiliou

Fifth Advisor

I-Jy Steven Chien

Abstract

Advances in wireless technology and portable computing along with demands for high user mobility have provided a major promotion toward the development of ad hoc networks. These networks feature dynamic topology, self-organization, limited bandwidth and battery power of a node. Unlike the existing commercial wireless systems and fixed infrastructure networks, they do not rely on specialized routers for path discovery and traffic routing. Security is an important issue in such networks. Typically, mobile nodes are significantly more susceptible to physical attacks than their wired counterparts.

This research intends to investigate the ad hoc network routing security by proposing a performance enhanced Secure ad hoc On-demand Routing protocol (SOR). Specifically, it presents a method to embed "Security Level" into ad hoc on-demand routing protocols using node-disjoint multipath, and to use "maximum hopcount" to restrict the number of routing packets in a specific area. The proposed scheme enables the use of security as a marked factor to improve the relevance of the routes discovered by ad hoc routing protocols. It provides customizable security to the flow of routing protocol messages. In general, SOR offers an alternative way to implement security in on-demand routing protocols.

Ad hoc network is too complex to allow analytical study for explicit performance expressions. This research presents a Stochastic Petri net-based approach to modeling and analysis of mobile ad hoc network. This work illustrates how this model is built as a scalable model and used to exploit the characteristics of the networks. The proposed scheme is a powerful analytical model that can be used to derive network performance much more easily than a simulation-based approach. Furthermore, the proposed model is extended to study the performance of ad hoc network security by adding multipath selection and security measurement parameters. This research gives a quantificational measurement to analyze the performance of a modified SPN model under the effect of multipath and attack of a hypothetical compromised node.

Share

COinS