Network Traffic Classification Using Machine Learning

Network traffic classification is crucial for managing and optimizing the performance of communication networks. Traditional traffic classification methods rely on port-based or payload-based approaches, which have become less effective due to the rise of encryption and dynamic port assignments. This paper proposes a classification methodology based on machine learning algorithms, to identify and categorize network traffic patterns. Then the identified patterns form a model capable of distinguishing “heavy traffic” connections from typical “light traffic” connections. The used Machine Learning algorithms include Decision Trees (DT), Support Vector Machines (SVMs), and K-Nearest Neighbors (KNN). The results of the proposed methodology demonstrate that machine learning algorithms significantly improve traffic classification accuracy, and the better evaluation criteria in predicting network traffic are the decision trees where DT accuracy is 94.71%, In contrast, the SVM accuracy is 91.52%, and KNN 94.28%.