Machine Learning Algorithms for Enhancing Intrusion Detection Within SDN/NFV

Emerging networks envisage to establish a modern digital society that tends to be more valuable on both social and economic levels. The objective is to resolve current network challenges and offer adequate security measures. As a consequence, adaptive architecture is necessary for upcoming networks. An emerging paradigm that can overcome the limitations of conventional networks is software-defined network (SDN), especially when coupled with Network Function Virtualization (NFV). It offers the capacity to dynamically manage and control the entire network by decoupling the control plane from the data plane. Nevertheless, various new network security issues must be handled. More opportunities to deliver intelligence inside of networks are given by SDN. This is why, thanks to SDN's characteristics, the use of machine learning methods is easily implemented. In this study, we introduce different existing network intrusion detection data sets, with a strong attention to SDN specific new dataset. Furthermore, we suggest an intelligent way to identify intrusions within SDN/NVF networks using a publicly available new SDN datasets (SDN Intrusion) and several Machine Learning techniques. Finally, we present and discuss our obtained results.