To Share or Not to Share? How Exploitation of Context Data Can Improve
In-Network QoE Monitoring of Encrypted YouTube Streams
I. Orsolic, L. Skorin-Kapov, and T. Hoßfeld. 2019 Eleventh International Conference on Quality of Multimedia Experience
(QoMEX) (QoMEX 2019), Berlin, Germany, (June 2019)
Abstract
With the widespread use of encryption in Over the Top (OTT) traffic,
Internet Service Providers (ISPs) for the most part lack insights into
application performance, as well as into Quality of Experience (QoE)
perceived by end users. Addressing challenges related to encryption, ISPs
are looking into machine learning based solutions that can detect
application performance solely from statistical properties of the traffic.
On the other hand, OTT service providers are not willing to share service
performance and content-related information with ISPs. While related work
on OTT-ISP collaboration scenarios has addressed architectural aspects,
business models, and to a certain extent incentives for sharing data, the
focus of this paper is on the exchanged data itself. We investigate to what
extent the performance of in-network machine learning based QoE estimation
models for HTTP adaptive video streaming could be improved with the
availability of certain context data provided by OTT providers. We motivate
OTT-ISP collaboration through more accurate in-network QoE monitoring and
potential improvement of user experience, which is of interest to both
sides.
%0 Conference Paper
%1 Orso1906:To
%A Orsolic, Irena
%A Skorin-Kapov, Lea
%A Hoßfeld, Tobias
%B 2019 Eleventh International Conference on Quality of Multimedia Experience
(QoMEX) (QoMEX 2019)
%C Berlin, Germany
%D 2019
%K QoE YouTube; encrypted learning machine monitoring; traffic;
%T To Share or Not to Share? How Exploitation of Context Data Can Improve
In-Network QoE Monitoring of Encrypted YouTube Streams
%X With the widespread use of encryption in Over the Top (OTT) traffic,
Internet Service Providers (ISPs) for the most part lack insights into
application performance, as well as into Quality of Experience (QoE)
perceived by end users. Addressing challenges related to encryption, ISPs
are looking into machine learning based solutions that can detect
application performance solely from statistical properties of the traffic.
On the other hand, OTT service providers are not willing to share service
performance and content-related information with ISPs. While related work
on OTT-ISP collaboration scenarios has addressed architectural aspects,
business models, and to a certain extent incentives for sharing data, the
focus of this paper is on the exchanged data itself. We investigate to what
extent the performance of in-network machine learning based QoE estimation
models for HTTP adaptive video streaming could be improved with the
availability of certain context data provided by OTT providers. We motivate
OTT-ISP collaboration through more accurate in-network QoE monitoring and
potential improvement of user experience, which is of interest to both
sides.
@inproceedings{Orso1906:To,
abstract = {With the widespread use of encryption in Over the Top (OTT) traffic,
Internet Service Providers (ISPs) for the most part lack insights into
application performance, as well as into Quality of Experience (QoE)
perceived by end users. Addressing challenges related to encryption, ISPs
are looking into machine learning based solutions that can detect
application performance solely from statistical properties of the traffic.
On the other hand, OTT service providers are not willing to share service
performance and content-related information with ISPs. While related work
on OTT-ISP collaboration scenarios has addressed architectural aspects,
business models, and to a certain extent incentives for sharing data, the
focus of this paper is on the exchanged data itself. We investigate to what
extent the performance of in-network machine learning based QoE estimation
models for HTTP adaptive video streaming could be improved with the
availability of certain context data provided by OTT providers. We motivate
OTT-ISP collaboration through more accurate in-network QoE monitoring and
potential improvement of user experience, which is of interest to both
sides.},
added-at = {2019-05-19T11:55:37.000+0200},
address = {Berlin, Germany},
author = {Orsolic, Irena and Skorin-Kapov, Lea and {Ho{\ss}feld}, Tobias},
biburl = {https://www.bibsonomy.org/bibtex/28ee8b2a2dadf5b95b9c130b0d2e82e13/hossfeld},
booktitle = {2019 Eleventh International Conference on Quality of Multimedia Experience
(QoMEX) (QoMEX 2019)},
days = {4},
interhash = {4cad29a87e05473bd132b9284cc3153c},
intrahash = {8ee8b2a2dadf5b95b9c130b0d2e82e13},
keywords = {QoE YouTube; encrypted learning machine monitoring; traffic;},
month = jun,
timestamp = {2019-05-19T11:55:37.000+0200},
title = {To Share or Not to Share? How Exploitation of Context Data Can Improve
{In-Network} {QoE} Monitoring of Encrypted {YouTube} Streams},
year = 2019
}