Analysis DDoS Attack Using Machine Learning On Software-Defined Network Architectures

Abstract

Software-Defined Network (SDN) architecture in the growing digital era is one of the solutions in improving efficiency, flexibility and scalability in managing computer networks. The concept of centralised control is the advantage of SDN architecture where a network administrator only forwards the rules made to all network devices. At the same time, this concept can also be a target for Distributed Denial of Service (DDoS) attacks. This research aims to classify traffic from SDN architecture using the MultiLayer Perceptron Classifier (MLPC) machine learning model to see whether the traffic is benign or malicious. The dataset used in this research is the ‘DDoS attack SDN Dataset’ which has a total of 23 features with a total of 104,345 data. The test results show that the MLPC model has an accuracy value of 99.05% and a precision and recall value of 99% in detecting benign and malicious traffic.