Models for evidence analysis for intrusion detection in ad-hoc networks.

摘要

Ad-Hoc Networks are a topic of considerable research interest due to their inherent advantages such as minimal infrastructure requirements, ease of setup and mobility. But with these advantages comes the disadvantages such as high security risks due to the wireless medium of communication, and the difficulty of monitoring the network to detect malicious behavior. In this thesis we enumerate some of these difficulties in intrusion detection. We then build upon an existing algorithm for intrusion detection and propose a new algorithm to incorporate the basic features from the existing solution. We propose to use Theory of Evidence to analyze the alerts from the IDS. The Dempster-Shafer Theory of Evidence is used to analyze the evidence and extended to account for the difference in confidence on different nodes reporting the alert.