Date of Award

8-25-2011

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Kenji Yoshigoe

Abstract

Protocol designers for wireless sensor networks (WSNs) have to consider many constraints including limited power and computing capacity of the sensing devices that make WSNs more vulnerable to attacks than wired or more resourceful wireless networks. The availability of a message for the adversary is the starting point for attacks on data confidentiality. This work proposes an end-to-end Message Dispersion Mechanism (MDM) to improve data confidentiality by significantly reducing message interception opportunities by potential adversaries in the first place. With MDM, a resource constrained sensor device first divides a message into several segments, encrypts them, and distributes them to a pre-defined set of super nodes (with significantly higher computing and communication capacity than sensor devices). Each super node then encrypts and forwards the segments to the super node set leader. Upon receiving all segments, the super node set leader reconstructs the original message. Both analytical and simulation evaluation show that MDM improves data confidentiality without adding significant communication and processing overhead. A further contribution of this study is the definition and derivation of a general expression for estimating the packet capture probability in a WSN. The model aids network design by considering parameters such as network size, required confidentiality level, budget for networking devices, traffic load, and adversary profile. It can be used along with network simulators to estimate the capture probability in the designed network.

Download

Share

COinS