Detecting P2P Botnets Network Traffic Behaviors Using Feature-Based Learning Techniques

Abstract

Botnets have become one of the major threats on the Internet. They are used to generate spam, carry out DDOS (Distributed Denial of Service) attacks and click-fraud, and steal sensitive information. Nowadays, many researchers interest to analyze the botnet technology and emphasis the botnet behaviors. It is needed to classify communication network traffic which is important fact to study the botnet behaviors. In this paper, we proposed an approach to detect botnet activity by analyzing and classifying network traffic behaviors due to P2P (Peer to Peer) based botnets. This system represents the important and most challenging types of botnet currently available that based on classifying P2P botnets. The classification techniques used in detection framework are RF (Random Forest) and SVM (Support Vector Machine). The performance evaluation of the two popular classification techniques is also presented. According to the experiments, proposed system has promising accuracy even with small time window by comparing two machine learning algorithms.