A Framework for in-Network QoE Monitoring of Encrypted Video Streaming

Abstract

With the amount of global network traffic steadily increasing, mainly due to video streaming services, network operators are faced with the challenge of efficiently managing their resources while meeting customer demands and expectations. A prerequisite for such Quality-of-Experience-driven (QoE) network traffic management is the monitoring and inference of application-level performance in terms of video Key Performance Indicators (KPIs) that directly influence end-user QoE. Given the persistent adoption of end-to-end encryption, operators lack direct insights into video quality metrics such as start-up delays, resolutions, or stalling events, which are needed to adequately estimate QoE and drive resource management decisions. Numerous solutions have been proposed to tackle this challenge on individual use-cases, most of them relying on machine learning (ML) for inferring KPIs from observable traffic patterns and statistics. In this paper, we summarize the key findings in state-of-the-art research on the topic. Going beyond previous work, we devise the concept of a generic framework for ML-based QoE/KPI monitoring of HTTP adaptive streaming (HAS) services, including model training, deployment, and re-evaluation. Components of the framework are designed in a generic way, independent of a particular streaming service and platform. The methodology for applying different framework components is discussed across various use-cases. In particular, we demonstrate framework applicability in a concrete use-case involving the YouTube service delivered to smartphones via the mobile YouTube app, as this presents one of the most prominent examples of accessing YouTube. We tackle both QoE/KPI estimation on a per-video-session level (utilizing the validated ITU-T P.1203 QoE model), as well as ''real-time'' KPI estimation over short time intervals. Obtained results provide important insights and challenges related to the deployment of a generic in-network QoE monitoring framework for encrypted video streams.INDEX TERMS Quality of Experience (QoE), video streaming, in-network QoE estimation, machine learning, encrypted traffic.

Measuring, Modeling and Integrating Time-Varying Video Quality in End-to-End Multimedia Service Delivery: A Review and Open Challenges

“…Therefore, future research works for the cross/multi-layer QoE monitoring solutions are needed. The collection of the QoE related KQIs from the customer premises/user-device/terminal also triggers concerns about user privacy and security which needs further research in the domain of the QoE monitoring [2], [226], [248]. There are few efforts in the state-of-the-art where QoE KQIs are extracted from the encrypted OTT video streaming session [248]- [253].…”

Uplink vs. Downlink: Machine Learning-Based Quality Prediction for HTTP Adaptive Video Streaming

“…Especially in recent years, several ML based approaches have been published for video QoE estimation. For YouTube in particular, according to Table 1 , an overall QoE prediction with real time approaches and focus on inspecting all network packets, is available from Mazhar [ 10 ], Wassermann [ 11 ], and Orsolic [ 12 ]. For all works, an in-depth analysis of voluminous network data is done, and more than 100 features each are extracted.…”

Crowdsourcing or Network KPIs? A Twofold Perspective for QoE Prediction in Cellular Networks

“…With similar aims, authors in [2] leveraged crowdsourcing-based LTE and WiFi network monitoring data to predict the quality of users' experience with respect to different applications, showing a 20% G-Mean improvement over baseline classifiers. Finally, it is worth mentioning that QoE prediction is an efficient driver for optimising resources allocation and enforcing maintenance strategies to meet customer demands and expectations [10], [11]. With this view, in [10] the authors design the concept of a generic framework for ML-based QoE/KPI monitoring of HTTP Adaptive Streaming (HAS) services, showing its applicability in a concrete test case of YouTube usage via customers smart-phones.…”

“…Finally, it is worth mentioning that QoE prediction is an efficient driver for optimising resources allocation and enforcing maintenance strategies to meet customer demands and expectations [10], [11]. With this view, in [10] the authors design the concept of a generic framework for ML-based QoE/KPI monitoring of HTTP Adaptive Streaming (HAS) services, showing its applicability in a concrete test case of YouTube usage via customers smart-phones. Also, in [11] the authors investigate the impact of QoE prediction errors in a crowdosurcing-based network monitoring system, obtaining insights which are generalizable and that provide interesting guidelines for network operators.…”