DDoS mitigation using blockchain and machine learning techniques

Online services are vulnerable to Distributed Denial of Service (DDoS) attacks, which overwhelm target servers with malicious traffic. These attacks are on the rise and challenging to detect due to their various forms, protocols, and the use of botnets. This paper presents a novel system that leverages machine learning algorithms for real-time DDoS attack detection and employs blockchain technology to store and block malicious IP addresses through software-defined networking. The system enhances security measures beyond traditional DDoS mitigation systems. In this paper, machine learning classification techniques are trained and tested using the Canadian Institute of Cyber Security's CICDDoS2019 dataset. Artificial Neural Network (ANN) outperformed KNN, Decision Tree, and Random Forest, achieving the best results. Additionally, the Ethereum blockchain is utilized to maintain a blacklist of malicious IP addresses. To assess the system's performance, a virtual network was established for testing using Mininet and the Python based Open-Source and OpenFlow (POX) controller. In real-time testing on the virtual network, ANN achieved an accuracy of 72.49%. This research presents a promising approach to combatting DDoS attacks while emphasizing the need for continuous improvement in cybersecurity.