Date on Master's Thesis/Doctoral Dissertation

5-2007

Document Type

Master's Thesis

Degree Name

Ph. D.

Department

Computer Engineering and Computer Science

Committee Chair

Kumar, Anup

Committee Co-Chair (if applicable)

Srinivasan, S.

Committee Member

Ragade, Rammohan K.

Committee Member

Kantardzic, Mehmed M.

Committee Member

Wong, Julius P.

Committee Member

Chang, Dar-Jen

Author's Keywords

MANET; Intrusion detection; Mobile ad hoc networks; Network security; Fuzzy logic

Subject

Computer networks--Security measures

Abstract

This dissertation presents a research whose objective is to design and develop an intrusion detection and response model for Mobile Ad hoc NETworks (MANET). Mobile ad hoc networks are infrastructure-free, pervasive and ubiquitous in nature, without any centralized authority. These unique MANET characteristics present several changes to secure them. The proposed security model is called the Intrusion Detection and Response for Mobile Ad hoc Networks (IDRMAN). The goal of the proposed model is to provide a security framework that will detect various attacks and take appropriate measures to control the attack automatically. This model is based on identifying critical system parameters of a MANET that are affected by various types of attacks, and continuously monitoring the values of these parameters to detect and respond to attacks. This dissertation explains the design and development of the detection framework and the response framework of the IDRMAN. The main aspects of the detection framework are data mining using CART to identify attack sensitive network parameters from the wealth of raw network data, statistical processing using six sigma to identify the thresholds for the attack sensitive parameters and quantification of the MANET node state through a measure called the Threat Index (TI) using fuzzy logic methodology. The main aspects of the response framework are intruder identification and intruder isolation through response action plans. The effectiveness of the detection and response framework is mathematically analyzed using probability techniques. The detection framework is also evaluated by performance comparison experiments with related models, and through performance evaluation experiments from scalability perspective. Performance metrics used for assessing the detection aspect of the proposed model are detection rate and false positive rate at different node mobility speed. Performance evaluation experiments for scalability are with respect to the size of the MANET, where more and more mobile nodes are added into the MANET at varied mobility speed. The results of both the mathematical analysis and the performance evaluation experiments demonstrate that the IDRMAN model is an effective and viable security model for MANET.

Share

COinS