Defending against malicious nodes using an SVM based Reputation System

Many networks, such as P2P networks, MANETs, file sharing networks, and online auction networks rely on node cooperation. If a malicious node gains access to such a network it can easily launch attacks, such as spreading viruses or spam, or attacking known vulnerabilities. Reputation systems (RS) defend against malicious nodes by observing their past behavior in order to predict their future behavior. These RSs usually comprise of statistical models or equations that are designed by hand and only defend against specific patterns of attacks. In this paper, we propose a support vector machines (SVM) based RS that defends against many patterns of attacks. It can be retrained to detect new attack patterns as well. We discuss the challenges associated with building RSs and how our RS tackles each of them. We compare the performance of our RS with another RS found in the literature, called TrustGuard, and perform detailed evaluations against a variety of attacks. The results show that our RS significantly outperforms TrustGuard. Even when the proportion of malicious nodes in the network is large, our RS can discriminate between good and malicious nodes with high accuracy. In addition our scheme has very low bandwidth overheads.