Machine Learning-Enabled Software-Defined Networks for QoE Management

Abstract

The use of artificial intelligence in modern technology is not a novelty. However, its use in telecommunication systems is little studied. This is not surprising because the implementation of third-party software modules in traditional networks is too complex, in contrast to software-defined networks, which have become very popular in the last few years. This paper presents the development of a machine learning module for predicting QoE (Quality of Experience) in software-defined networks. The module allows to predict and provide customer-defined quality of service and reduce the load on network equipment by reducing the amount of signaling traffic in the network. A software-defined network architecture with an integrated machine learning module is proposed. The study compared the predicted QoE level and the results obtained with the QoE monitoring system in the virtual network in the Mininet environment. In addition, graphs of the network load by signal packets using the machine learning module and the standard monitoring system are presented. We have proven that using a machine learning module reduces signaling traffic on the network by 30%.