Let me Decrypt your Beauty: Real-time Prediction of Video Resolution and Bitrate for Encrypted Video Streaming

The dynamic adaptation of the video quality induced by HTTP Adaptive Streaming (HAS) technology introduces new Quality of Experience (QoE) metrics beyond re-buffering. In this work we address the problem of real-time QoE monitoring of HAS, focusing on the continuous prediction of video resolution and average video bitrate, for the particular case of YouTube. Through empirical evaluations over a large video dataset, we demonstrate that it is possible to accurately predict the specific video resolution, as well as the average video bitrate, both in real time, and using a time granularity as small as one new prediction every second, not achieved by other proposals in the literature.

[1]  Michael Seufert,et al.  Features that Matter: Feature Selection for On-line Stalling Prediction in Encrypted Video Streaming , 2019, IEEE INFOCOM 2019 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS).

[2]  Li Gang,et al.  Stream-based Machine Learning for Real-time QoE Analysis of Encrypted Video Streaming Traffic , 2019, 2019 22nd Conference on Innovation in Clouds, Internet and Networks and Workshops (ICIN).

[3]  Michael Seufert,et al.  A Public Dataset for YouTube's Mobile Streaming Client , 2018, 2018 Network Traffic Measurement and Analysis Conference (TMA).

[4]  Lea Skorin-Kapov,et al.  YouTube QoE Estimation Based on the Analysis of Encrypted Network Traffic Using Machine Learning , 2016, 2016 IEEE Globecom Workshops (GC Wkshps).

[5]  Katherine Guo,et al.  Requet: real-time QoE detection for encrypted YouTube traffic , 2019, MMSys.

[6]  Zubair Shafiq,et al.  Real-time Video Quality of Experience Monitoring for HTTPS and QUIC , 2018, IEEE INFOCOM 2018 - IEEE Conference on Computer Communications.