SVM-based detection of ad hoc flooding attacks

In recent years ad hoc networks have become very attractive for many applications such as tactical and disaster recovery operations. However they are vulnerable to many attacks. The vulnerabilities of wired networks such as denial of service (DoS), eavesdropping, spoofing and the like, becomes more acute in these networks. Especially it is hard to differentiate DoS attacks in these highly dynamic systems. In this research, we design an intrusion detection model using Support Vector Machines (SVM) in order to detect a popular DoS attack on these networks, namely ad hoc flooding attacks. We evaluate its performance on simulated networks with varying traffic and mobility patterns.