A Review on Software Defined Networking-Based Network Traffic Classification Using Machine Learning Techniques

With the rise in internet usage, accurate network traffic classification has become essential for secure and efficient communication. Traditional methods like port-based and payload inspection are no longer reliable due to dynamic traffic patterns and growing encryption. To address these challenges, this study applies Machine Learning (ML) models within a Software Defined Networking (SDN) environment to classify DNS, Telnet, Ping, and Voice traffic. Using D-ITG for traffic generation and Mininet for SDN simulation, multiple ML algorithms including Logistic Regression, Decision Tree, Random Forest, AdaBoost, SVM, and K-means were evaluated after thorough preprocessing and feature extraction. Among them, the Decision Tree achieved the highest accuracy of 99.81%. Overall, the findings show that ML integrated with SDN significantly enhances offline and real-time traffic classification, even for encrypted traffic, while improving QoS.