AI/ML-based real-time classification of Software Defined Networking traffic

Abstract

One particular example of a useful software application for Software Defined Networks (SDN) is represented by a traffic analysis mechanism, which provides a network administrator with a control panel from which he can collect traffic data. The data can then be used to fit Artificial Intelligence (AI) models, which will further classify the traffic of the network in real-time, enabling a network admin to monitor the network with ease. This paper presents an SDN classifier, aiming to achieve real-time multi-class traffic classification in a software-defined network. To enhance the classification accuracy, six artificial intelligence algorithms, including Logistic Regression, K-Nearest Neighbors (KNN), Naïve Bayes, Support Vector Machines (SVM), Decision Tree, and Artificial Neural Networks (ANN), are tested. Due to the possibility of training on unnormalized data, the data is preprocessed by rescaling values between 0 and 1. Additionally, the paper explores the supervised learning potential of the last three algorithms in traffic classification. The findings show that one of the top performing algorithms is ANN, along with SVM and KNN.