5G aims to provide a complete wireless communication system with various applications, network services and technologies. In terms of 5G network management, Software-Defined Networking (SDN), and Network Functions Virtualization (NFV)are expected to control and manage network resources. Network Softwarization provides better management of network traffic. However, it does not guarantee network performance will not degradation when the traffic rises. Flow identification has been raised as a solution for keeping the network performance, and it has become a hot topic in both, academy and industry. In particular, there is a high interest in identifying video streaming flows since thanks to 5G and its benefits that improve the streaming media industry, the video streaming traffic is expected to increase dramatically due to the massive connection of 5G compatible devices. Motivated by this, we presented a novel approach for identifying video streaming services. Our approach includes three modules: video stream acquisition module, video stream analyzer module, and application module. In the video stream acquisition module, we capture video streaming packets and organize them in to flow records. In the video streaming analyzer module, we analyze the flow records using supervised machine learning algorithms to find the appropriate algorithm that performs better. In the application module, we provide a brief explanation of the applications of our approach. Additionally, we provide an analysis of the overall workload generated by our approach. The results of the evaluation by module corroborate the usefulness and feasibility of our approach for identifying video streaming services.

中文翻译：

---

基于知识定义网络的 5G 网络视频流识别方法


5G旨在提供一个包含各种应用、网络服务和技术的完整无线通信系统。在5G网络管理方面，软件定义网络（SDN）和网络功能虚拟化（NFV）有望控制和管理网络资源。网络软件化提供了更好的网络流量管理。但它并不能保证网络性能在流量上升时不会下降。流识别作为一种保持网络性能的解决方案而被提出，并成为学术界和工业界的研究热点。特别是，由于 5G 及其改善流媒体行业的优势，人们对识别视频流流量非常感兴趣，由于 5G 兼容设备的大量连接，预计视频流流量将大幅增加。受此启发，我们提出了一种识别视频流服务的新颖方法。我们的方法包括三个模块：视频流采集模块、视频流分析器模块和应用模块。在视频流采集模块中，我们捕获视频流数据包并将它们组织成流记录。在视频流分析器模块中，我们使用监督机器学习算法分析流记录，以找到性能更好的合适算法。在应用模块中，我们对我们的方法的应用进行了简要说明。此外，我们还对我们的方法生成的总体工作负载进行了分析。模块评估的结果证实了我们识别视频流服务的方法的有用性和可行性。