direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

QoE modeling for complex service scenarios: MMORPGs and OTT video streaming

LOCATION:  TEL, Room 209 (2nd floor), Ernst-Reuter-Platz 7, 10587 Berlin

Date/Time: 09.04.2018, 14:15-15:00  

SPEAKER: Mirko Sužnjević (University of Zagreb)

ABSTRACT:

The first part of this talk will focus on methodologies for assessing and modeling QoE for multiplayer games. The talk presents the results of a measurement study involving one of the most popular Massively Multiplayer Online Role-Playing Games (MMORPGs) – World of Warcraft, with more than 100 participants taking part in QoE assessment. MMORPGs present very complex game systems with a multitude of different gameplay elements. System, user, and context parameters were measured and analyzed in detail in the study. Other QoE metrics besides MOS were investigated, but the modeling procedure was done only for MOS. Both linear and machine learning models will be presented.

The second part of the talk will focus on challenges related to monitoring and estimating QoE in video streaming services such as YouTube. Due to privacy issues, most streaming services have now introduced traffic encryption, thus deeming DPI methods for reading QoE-relevant data from headers obsolete. Today, ISPs for the most part rely solely on the monitoring of statistical traffic characteristics for QoE/KPI estimation and root cause analysis of potential quality problems. Popular video services implement the HTTP Adaptive Streaming (HAS) paradigm, which makes this problem even more challenging. In the talk our machine learning approach for estimating QoE of YouTube video streams solely from network traffic traces will be described, as well with the results for different platforms (Android and iOS) and access networks (mobile and WiFi).

 

 

Zusatzinformationen / Extras

Direktzugang

Schnellnavigation zur Seite über Nummerneingabe