Lightweight Machine Learning Prediction Algorithm for Network Attack on Software Defined Network

Nowadays, Software Defined Networking (SDN) technology is massively applied in network infrastructures. However, the SDN has several vulnerabilities to various network attacks, such as Denial of Service (DOS) for attacking the availability of the network and various web-based attacks, such as brute force and privilege escalation attacks. This research proposes a Machine Learning method to classify malicious traffic. The InSDN dataset will represent malicious traffic in the SDN environment. Feature correlation is used in this research to reduce the number of the features in InSDN dataset. The reduced dataset feature gives the fastest learning time with respect to the original dataset. The random forest gives the best metric with 99.9962% in accuracy with respect to learning methods, such as KNN and decision tree.