A Hybrid Machine Learning Approach for the Detection and Prevention of DDoS Attacks in Software-Defined Networks

Software-Defined Networks (SDN) have revolutionized the computer network industry by providing software-based pro-grammable network controllers for communicating with the hardware infrastructure and controlling the traffic on the network. In SDN, a virtual network can be created using virtual network devices, enabling efficient control of network traffic and monitoring of traditional network devices. While SDN offers a great deal of adaptability, it is also susceptible to attacks such as Distributed Denial of Service (DDoS), which can bring down the whole network very quickly. This study proposes a machine learning technique to distinguish between normal and DDoS attack traffic on SDN. The main contribution of this research is the discovery of new features for detecting DDoS attacks from the SDN traffic dataset. Classification is carried out using a novel hybrid machine-learning approach. Based on the findings, the traffic is most accurately classified by the proposed hybrid model combining XGB Classifier and Random Forest (RF), which achieves a 99.11% accuracy for distinguishing between benign and attack network traffic. Our findings are consistent with previous results showing that the accuracy of the Decision Tree classifier is 83.79%, the accuracy of the Random Forest classifier is 98.24% and the accuracy of the XGB classifier is 98.00% which also represents the best performance.