Detection of Misuse Attack in NFV Networks Using Machine Learning

Abstract

Network Function Virtualization (NFV) represents a virtual network whose service is provided by virtual parts of virtual machines. This type of network is easy to implement and update. In addition, NFV leading to low cost due to sharing the same resources. As is the case with other networks, NFV is not safe from attacks. Since all parts of this NFV network share the same resources, misuse attack is regarded to be the most common attack in NFVs, particularly because the attack use one or more of the resources which affect all parts of the NFV. This paper is based on using machine learning to extract rules of misuse attack detections. The tree decision C4.5 algorithm has been used to extract these rules, with nine features of network data flow. When testing the propose work with a server traffic data having more than 5 million network connections, the results show that a comparatively higher performance of the algorithm C4.5 with an accuracy of about 96%.