Research useful and service traffic based on machine learning in multiservice software defined network

Traffic classifications based on digital end-to-end technologies such as Software Defined Networking, Network Functions Virtualization and Machine Learning have been studied with an increase in the volume of transmitted traffic and in the provision infocommunication services.The analyzes showed that critical information and communication services require proactive monitoring of the Service Level Agreement in real time and guaranteed quality of service, which must be tested and monitored during the transmission, processing and reception traffic packet flows.This paper investigates the problems dynamic classification packet flows useful and service traffic using machine learning in a multiservice software-defined network.Methods for calculating dynamic traffic classification based on machine learning using intelligent technology are proposed and analytical expressions are used to evaluate the quality metrics of the classifier's work, such as overall accuracy, accuracy and completeness each class, and F1-measure using the Random Forest and XGBoost methods.