Detecting DDoS Attacks in Software-Defined Networking through the Utilization of Supervised and Unsupervised Learning Techniques
T. Sathwik, Tadi Sai Vishruth Reddy, Venkat Rohith Gangavarapu, Sreebha Bhaskaran · 2024
Given the dynamic nature of SDN environments, it is necessary to quickly recognize Distributed Denial of Service (DDoS) attacks to maintain our network’s integrity and ensure uninterrupted service delivery. However, the centralized control plane of SDN makes it particularly vulnerable, as a compromise here can allow attackers to manipulate network traffic and configurations extensively. The traditional detection methods often have trouble keeping up with these attacks’ dynamic nature, resulting in high false positive rates and poor response. This paper proposes an approach that uses supervised and unsupervised learning techniques to enhance DDoS attack detection capabilities in SDN infrastructures. Using Mininet and Ryu, our method involves collecting network traffic data, extracting relevant features, and constructing classification models based on labelled datasets that can identify typical attack patterns. Additionally, these initiatives are supported by unsupervised learning approaches like clustering or anomaly detection, which help reveal new types of attack signatures not seen before and abnormal behaviours within networks caused by them. As a result, what we have done here enables real-time detection against DDoS through seamless integration into the Ryu controller, thus strengthening the security posture against emerging threats in cyberspace targeting SDNs.