Machine Learning Techniques for Intrusion Detection in SDN: A Survey

Abstract

AbstractThe SDN (Software Defined Network) was developed to reduce the complexity of the traditional network by controlling and managing the entire network from a centralized location. This new model introduces the separation of the transmission plane and the control plane by providing a new independent plane called the SDN controller. Today, due to the various benefits of this new paradigm, many commercial and industrial enterprises are converging to this solution in their network environments, especially in data centers. Nevertheless, the emerging SDN technology can lead to many security vulnerabilities and threats such as man-in-the-middle (MITM) attacks, denial-of-service (DoS), overload saturation attacks. Therefore, deploying Intrusion Detection Systems (IDS) to monitor malicious activities is crucial for the SDN network architecture. This paper aims to review the different attacks and intrusions affecting the SDN environment through the various existing studies conducted in recent years by the research communities. Then, we will study the different approaches or machine learning techniques such as Naive Bayesian Classification, Support Vector Machine (SVM), etc., to classify and detect anomalies and intrusions in an SDN architecture to choose the most appropriate and efficient algorithm.KeywordsSDNMachine learningIntrusion detectionSoftware-defined network