Implementation of Machine Learning-Based QoS Traffic Classification In Software-Defined Networking

The concept of Software-Defined Networking (SDN) in network management involves controlling network behaviour through software applications. The implementation of SDN centralizes network management, ensuring efficient performance and reducing computational costs. With the vast amount of network traffic today, traffic classification is crucial in SDN, improving performance by prioritizing applications that require higher Quality of Service (QoS). The occurrence of imbalanced datasets, which focus more on the majority class rather than the minority class, leads to poor performance in SDN. This paper explores the implementation of feature selection using filter and wrapper methods to tackle imbalanced datasets and enhance network management performance. Using a real-time dataset containing 930 features from the University of Cambridge, the Symmetrical Uncertainty and C4.5 algorithm achieve a classification accuracy of 97.69% and reduced the dataset size by 50%. These findings demonstrate that the proposed methods significantly improve computational efficiency and classification performance, making them suitable for real-time SDN applications.