Flow Based Intrusion Detection System for Software Defined Networking using Hybrid Machine Learning Technique

Abstract

Software Defined Networking and OpenFlow protocol have been recently emerged as dynamic and promising framework for future networks. Even though, programmable features and logically centralized controller leads to large number of security issues. To address the security problems, we have to impose Intrusion Detection System module to continuously keep track of the network traffic and to detect the malicious activities in the SDN environment. In this paper, we have implemented flow-based IDS with the help of hybrid machine learning technique. By collecting the flow information from the controller, we classify the traffic, extract the essential features and classify the attack using machine learning based classifier module. For classifier, we have developed hybrid machine learning model with the help of Modified K-Means and C4.5 algorithm. Our proposed work is compared with single machine learning classifier and our experimental results show that, proposed work can classify the normal and attack instances with accuracy of 97.66%.