Download now Free registration required
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, the authors propose a Support Vector Machines (SVM) based RS that defends against many patterns of attacks.
- Format: PDF
- Size: 616 KB