ANOMALY DETECTION USING MACHINE LEARNING ON SOFTWARE DEFINED NETWORKING

Abstract: Software-defined networking (SDN) has experienced significant growth and can be leveraged across various network scenarios, ranging from data centres to wide-area 5G networks. It transfers control logic from individual devices to a centralized programmable controller, enabling efficient monitoring and management of network traffic. While a software-based controller enforces rules and policies on forwarded requests, it lacks the ability to identify abnormal patterns in network traffic. Consequently, the controller may inadvertently install flow rules that counteract these anomalies, resulting in reduced overall network performance. These anomalies could indicate potential threats to the network, thereby compromising its security and performance. To address this, machine learning (ML) approaches can be employed to detect such traffic flow patterns and anticipate impending threats to the system. In this study, we propose an ML-based system for detecting traffic anomalies in software-defined networks, specifically utilizing the Support Vector Machine (SVM) algorithm for anomaly detection.

Keywords: Software-defined networking, Abnormal patterns in network traffic, Machine learning, Support Vector Machine algorithm.