MACHINE LEARNING-BASED PATH SELECTION DEVICE AND METHOD IN SOFTWARE DEFINED NETWORK (SDN) ENVIRONMENT

摘要

The present invention relates to a machine learning-based path selection device and method in a software defined network (SDN) environment. According to the present invention, the path selection method receives state information including port information and flow information collected from each of a plurality of switches from an SDN controller, and traffic throughput of each port of the plurality of switches, and trains a traffic throughput prediction model to output traffic throughput for each port when the state information is input. Then, the method inputs current state information of a plurality of paths capable of transmitting a packet from a first terminal to a second terminal and the plurality of switches located in the plurality of paths in the traffic throughput prediction model, and selects any one of the plurality of paths.;COPYRIGHT KIPO 2020