Machine Learning for Network Traffic Classification in Software-Defined Networks

Authors

Perumallapalli Ravikumar

Abstract

Software-Defined Networks' (SDNs') quick development has given network administration previously unheard-of flexibility and programmability. But there are also serious difficulties in efficiently controlling and safeguarding network traffic because of this flexibility. Due to the dynamic and complicated nature of contemporary network environments, traditional traffic categorization techniques—which mostly rely on predetermined rules and signatures—are frequently insufficient. The implementation of machine learning techniques for network traffic classification within SDNs is examined in this research in order to overcome these issues. The main goal is to improve overall network performance and security by using cutting-edge machine learning models to increase the efficiency and accuracy of traffic classification. A thorough literature review is used to determine the advantages and disadvantages of current approaches. In order to classify network traffic, we then provide a novel framework that makes use of deep neural networks (DNNs). In order to manage high-dimensional traffic data and adjust to evolving traffic patterns, our approach does not require large amounts of labeled datasets. With a 95% classification accuracy, the suggested approach was tested on both real-world and simulated SDN traffic data. Furthermore, as compared to conventional techniques, it showed notable gains in computing efficiency, recall, and precision. The benefits and drawbacks of the present methods are assessed through a comprehensive literature study. We then provide a new framework that uses deep neural networks (DNNs) to classify network traffic. Our method doesn't require a lot of labeled datasets to handle high-dimensional traffic data and adapt to changing traffic patterns. The proposed method was evaluated on simulated and real-world SDN traffic data and achieved 95% classification accuracy. Moreover, it demonstrated significant improvements in computing efficiency, recall, and precision when compared to traditional methods.

Keywords

Machine learning, SDNs, network traffic classification, deep neural networks, network management, security, QoS.

Machine Learning for Network Traffic Classification in Software-Defined Networks. Perumallapalli Ravikumar. 2017. IJIRCT, Volume 3, Issue 1. Pages 1-9. https://www.ijirct.org/viewPaper.php?paperId=2412037