Enhanced Machine Learning-Based SDN Controller Framework for Securing IoT Networks

Abstract

The Internet of Things (IoT) ecosystem consists of interconnected devices that work together. It facilitates communication between devices, real-time data exchange and cloud computing. The occurrence risk of cyber attacks grows exponentially with these interconnected systems. Devices, networks and data could easily come under attack, wherefore they become vulnerable and could be compromised by hackers. To address this problem, we propose an enhanced Framework used on Software Defined Network (SDN) environment-based Intrusion Detection System (IDS) for securing IoT devices from malicious activity. We implement the machine learning (ML) method as part of the SDN controller's Network Intrusion Detection System (NIDS). Our enhanced ML-based SDN Controller Framework (Improved ML-SDN) classify the data traffic and makes a real-time prediction. It is based on K-Nearest Neighbor (kNN) supervised learning algorithm with others improving model Accuracy. It has produced an accuracy of 99.7%, 0.02304 s of building model time, 0.2997 s of detection time and a false alarm rate (FAR) of 0.34%.