Learning-based Orchestrator for Intelligent Software-defined Networking Controllers

Abstract

This paper presents an overview of our learning-based orchestrator for intelligent Open vSwitch that we present this using Machine Learning in Software-Defined Networking technology. The first task consists of extracting relevant information from the Data flow generated from a SDN and using them to learn, to predict and to accurately identify the optimal destination OVS using Reinforcement Learning and QLearning Algorithm. The second task consists to select this using our hybrid orchestrator the optimal Intelligent SDN controllers with Supervised Learning. Therefore, we propose as a solution using Intelligent Software-Defined Networking controllers (SDN) frameworks, OpenFlow deployments and a new intelligent hybrid Orchestration for multi SDN controllers. After that, we feeded these feature to a Convolutional Neural Network model to separate the classes that we’re working on. The result was very promising the model achieved an accuracy of 72.7% on a database of 16 classes. In any case, this paper sheds light to researchers looking for the trade-offs between SDN performance and IA customization.