A comprehensive overview of machine learning for intrusion detection in software-defined networking

A comprehensive overview of machine learning for intrusion detection in software-defined networking

연구 분야: Safety

학회: Innovations in Systems and Software Engineering

In Software-Defined Networking (SDN), the separation of control and data planes enables centralized management through a single controller, which calculates and relays forwarding rules to the data plane via a control protocol. While SDN architecture offers enhanced flexibility, programmability, and centralized oversight, it also introduces significant security vulnerabilities that expose SDN-based systems to various attacks. This study explores the security challenges inherent in SDN architectures, focusing on the implementation of Intrusion Detection Systems (IDS) that utilize machine learning techniques to detect network attacks effectively within SDN environments. Our research investigates the effectiveness of different machine learning models tailored to specific attack types, emphasizing the necessity of aligning model selection with the unique characteristics of each attack. Through a comparative analysis, we evaluate the performance of these models, highlighting critical factors such as detection accuracy, computational overhead, and feature selection.