Research on the traffic type recognition technique for advanced network control using Floodlight

摘要

Due to the development of network technology and the recent success of 5G network commercialization, research on SDN and its application are once again increasing. SDN / NFV has created an environment in which various methods can be studied, avoiding the hardware dependencies that were structural limitations. In recent years, many researches are conducted to integrate machine learning technology into the network field and technologies such as traffic control, congestion control and efficient routing algorithms. According to the requirements of 5G network such as eMBB (enhanced Mobile Broadband), mMTC(massive machine type communications), and URLLC(Ultra-Reliable and Low Latency Communication), more and more various packets are actually appearing on the network, growing need for network management techniques to handle these various packets. Especially, with steep rise of the number of application service having distinct characteristics such as IoT, VR and 4K video, it is obvious that there is a limitation to use previous general-purpose network management techniques for them. Therefore, there is a need for research and development of adaptive networking technology which classify usage and types of network packet. Several existing machine learning algorithms have been used to classify protocols of target packets and it made a good result in previous researches. In contrast, this paper more focuses on the usage and types of packets and verify accuracy and performance by applying clustering algorithm using Deep Autoencoder. This is expected to enable optimized network management for each application and provide basic research base and experimental results of more advanced network management technology.