Degradation Detection of Video Streaming on Mobile and Fixed Networks Using Digital Twins

使用应用和网络级度量来预测视频流的体验质量（QoE）已经取得了巨大的成功。其中大部分都是通过模拟不同网络条件下的视频流会话并记录视频的质量或观众对流媒体视频的感知来完成的。本文探讨了数据驱动的性能数字孪生（DT）的使用，DT将作为本研究的一部分，作为一个平台，可以使用AI/ML对QUIC和TCP协议上的视频流进行降级检测（人工智能/机器学习）和建议是什么导致了降级，以便网络运营商可以使用它们来优化他们的网络。YouTube视频将用于这项研究。（摘要摘自2022年8月的第1年进展报告。将于2023年底前更新）

There has been tremendous success with predicting quality of experience (QoE) of video streaming using application and network level metrics. Most of these have been done by simulating video streaming session over different network conditions and recording the video's quality or viewers' perception of the streamed video.This paper explores the use of Data-driven Performance Digital Twin (DT) which will be built as part of this research, as a platform on which degradation detection of video streaming over QUIC and TCP protocols can be carried out using AI/ML (Artificial Intelligence/Machine Learning) and suggestion of what caused the degradation so that network operators can use them to optimize their network. YouTube videos will be used in this research.(Abstract taken from Year 1 Progression Report, August 2022. It will be updated by the end of 2023)