Detection and Classification of Distributed DoS Attacks Using Machine Learning

Distributed denial of service (DDoS) attacks target the websites and online services in which the attacker floods them with more traffic than the server or network that can hold. With the sophistication of technology and the rise of the Internet, such attacks are becoming more common and easier to perform. According to a study by the Cisco Visual Networking Index (VNI) in 2017, global projections of the overall number of DDoS attacks would significantly increase by 2022 and might cross 14.5 million marks, thereby doubling in number. A system is proposed that aims to provide an efficient way for the detection of DDoS attacks in a network. As previously, machine learning has been widely used for intrusion detection and classification of the type of attack compared to other techniques and intrusion detection system (IDS). This system makes use of different machine learning algorithms (extreme gradient boosting, K-nearest neighbour, stochastic gradient descent, and Naive Bayes) and a deep learning architecture (convoluted neural network) to identify attacks and classify them. The result shows that XGBoost achieves the highest accuracy, while CNN and KNN also give comparable figures. Our code is available at https://​github.​com/​mohak1/​Detection-and-Classification-of-Distributed-DoS-Attacks-using-Machine-Learning.