Soutenance de thèse de doctorat de M. Mamadou Tourad DIALLO . Cette thèse a été réalisée sous la direction de Mr Hossam AFIFI. Elle aura lieu le 04 Juin 2015 à 14 H, Batiment 862 à l’Amphithéâtre 34, Saclay Nano Innov, 8 Avenue de la vauve, 91120 Palaiseau.
Le jury sera composé de :
Adlen KSENTINI, Professeur Associé, Univ. de Rennes 1, France, Rapporteur
Ken CHEN, Professeur, Univ. de Paris 13, France, Rapporteur
Hassnaa MOUSTAFA, Senior Scientist, Intel Corporation, USA, Examinateur
Luigi ATZORI, Professeur, Univ. de Cagliari, Italie, Examinateur
Laurent RUCKENBUSCH, Team Manager, Orange, France, Examinateur
Houda LABIOD, Professeur Associé, Telecom ParisTech, France, Examinateur
Hossam AFIFI, Professeur, Telecom SudParis, France, Directeur de thèse
Nicolas MARECHAL, Project Leader, Orange, France, Co-directeur de thèse
Abstract :
Video services are experiencing an unprecedented growth for the last few years. As such, market players are fighting to increase Average Revenue per User (ARPU), limit churn and improve their market share. From a marketing point of view, one possible option is to focus on improving end-users’ satisfaction, namely Quality of Experience (QoE). QoE is a young research subject with limited considerations regarding contextual information, that deserves a deeper understanding and can offer great commercial perspectives.
Semantically speaking, this concept is closely related to the Quality of Service (QoS), even if the former is now associated to the enforcement of purely technical constraints so that to ensure a given level of service expectations. Contrary to this, QoE goes beyond the technical background.
In this PhD thesis, we first provide a technical overview on video services and architectural deployments for IPTV (Internet Protocol TeleVision) and WebTV (Web TeleVision) services. Then, a state-of-the-art about both QoE measurement techniques and Content & Delivery Adaptation is also provided. According to these surveys, two methods can be considered to understand how users interact with services and estimate their QoE. On one hand, by monitoring and analyzing the impact of quality metrics on user engagement, in order to understand the effects of technical video metrics (video startup time, average bitrate, buffering ratio) and content popularity on user engagement. On other hand, one can consider subjective approaches such as the Mean Opinion Score (MOS) for evaluating QoE, in which users are required to give their assessment/rating. The MOS presents the exact user perception of the viewed video, which is considered as a better indicator of video quality as it is given by humans.
Our results show that video the buffering ratio and content popularity are critical parameters which strongly impacts the end-user’s satisfaction and user engagement, while the video startup time appears as less significant. In the third part, we propose to assess QoE in terms of MOS (Mean Opinion Score) through introducing contextual information. We did tests with users to get their feelings while watching video contents under varying conditions (context parameters). A detailed
overview and statistical analysis of our study shows the existence of non-trivial parameters impacting MOS (the type of device, the content type for constant video bitrate: football, cartoon etc.). We also propose mathematical models to develop functional relationships between the QoE and the context information which in permits us to estimate the QoE. To assess the performance of our proposal, we compare it with an operational QoE measurement tool. Our results prove that contextual information is an important parameter and one which needs to be taken into account for monitoring and providing an accurate assessment of QoE. Finally, in the last part of this manuscript, we provide general QoE multi-users optimization, that optimizes the network resources by considering the end-user satisfaction in terms of MOS. Our proposals improve the perceived QoE for different video sessions sharing the same local network, while taking QoE fairness among users as a leitmotiv. This approach is validated by simulations and corresponding prototype architecture is proposed.
In conclusion, the different contributions proposed in this thesis improves the understanding of hidden relationships between quality parameters and user engagement, and about how contextual information may inuence the end-user’s perceived video quality. Finally, we proved that this work can help improving the network usage, reducing congestion phenomena and in ensuring a level of QoE for connected users.
Keywords: Multimedia, Audio-Visual Services, Video Streaming, Content Adaptation, QoE, Contextual Information