Detection of DDoS Attacks using Machine Learning Algorithms

Abstract

Distributed Denial of Service attack (DDoS) is the most dangerous attack in the field of network security. DDoS attack halts normal functionality of critical services of various online applications. Systems under DDoS attacks remain busy with false requests (Bots) rather than providing services to legitimate users. These attacks are increasing day by day and have become more and more sophisticated. So, it has become difficult to detect these attacks and secure online services from these attacks. In this paper, we have used machine learning based approach to detect and classify different types of network traffic flows. The proposed approach is validated using a new dataset which is having mixture of various modern types of attacks such as HTTP flood, SID DoS and normal traffic. A machine learning tool called WEKA is used to classify various types of attacks. It has been observed that J48 algorithm produced best results as compared to Random Forest and Naïve Bayes algorithms.