Signalprint-Based Intrusion Detection in Wireless Networks

Wireless networks are a critical part of global comnninication for which intrusion detection techniques should be applied to secure network access, or the cost associated with successful attacks will overshadow the benefits that wireless networks offer. In this paper we investigate a new scheme called Nodeprints to extend the existing centralized Signalprints design for authentication to a distributed voting-based design for intrusion detection. We analyze the effect of voting-based intrusion detection designs, the probability of an individual node voting incorrectly, the ratio of mobile nodes to base stations, and the rate at which nodes are compromised, on the system performance measured by the probability that the intrusion detection system yields a false result. We develop a performance model for evaluating our Nodeprints design and identify conditions under which Nodeprints outperforms the existing Signalprints design.