Machine learning in SDN


Volkov S.


Increase in demand for network connectivity has challenged traditional network architectures. To match demand, SDN (Software-Defined Network) was proposed as a new architecture. Since SDN technology provides network virtualization capabilities, separates control and data planes, implements logically centralized control and opens up network capabilities for higher-level applications, it is especially suitable for implementing data center networks. This network will be distinguished by functionality that supports centralized management. This article provides an overview of software-defined network technology. The features of the architecture of these networks are described, as well as the main advantages of this technology over the architecture of traditional networks. The issue of security in the SDN is considered. The authors concluded that it is possible to solve the security problem of software-defined networks using machine learning methods. A review of various studies and experiments on the use of these methods to detect and prevent potential attacks in the SDN is presented. Machine learning methods also can be used to analyze traffic taking into account QoS (Quality of Service). Several works on ensuring the quality of service for software-defined networks are considered. Among them there are works that also use machine learning methods.

External links


At the International Journal of Open Information Technologies website:

Download PDF or read online at CyberLeninka (in Russian):

RUDN University. Repository:

Reference link

S. S. Volkov, I. I. Kurochkin. Machine learning in SDN // International Journal of Open Information Technologies. – 2019. – Vol. 7. – № 11. – Pp. 49-58.