Jérémie Leguay s’est rendu à l’IRT SystemX le 14 février pour animer un séminaire sur le thème « Optimisation de la qualité d’expérience avec les réseaux logiciels ».
Résumé (en anglais)
The surge of video traffic is a challenge for service providers that need to maximize Quality of Experience (QoE) while optimizing the cost of their infrastructure. Classic network control techniques have as sole objective the fulfillment of Quality-of-Service (QoS) metrics, being quantitative and network-centric. Nowadays, the research community envisions a paradigm shift that will put the emphasis on Quality of Experience (QoE) metrics, which relate directly to the user satisfaction. Yet, assessing QoE from QoS measurements is a challenging task that powerful Software Defined Network controllers are now able to tackle via optimization and machine learning techniques.
In the first part, we address the problem of routing multiple HTTP-based Adaptive Streaming (HAS) sessions to maximize QoE. We design a QoS-QoE model incorporating different QoE metrics which is able to learn online network variations and predict their impact on representative classes of adaptation logic, video motion and client resolution. Different QoE metrics are then combined into a QoE score based on ITU-T Rec. P.1202.2. This rich score is used to formulate the routing problem. We show that, even with a piece-wise linear QoE function in the objective, the routing problem without controlled rate allocation is non-linear. We therefore express a routing-plus-rate allocation problem and make it scalable with a dual subgradient approach based on Lagrangian relaxation where sub problems select a single path for each request with a trivial search, thereby connecting explicitly QoE, QoE and HAS bitrate. We show with ns-3 simulations that our algorithm provides values for HAS QoE metrics (quality, rebufferings, variation) equivalent to MILP and better than QoS-based approaches.
Then, in the second part, we focus on a few crucial QoE factors and propose machine learning models to predict re-buffering ratio, average video bitrate and quality variations. We show that hidden variable models based and context information that a network controller could provide can boost accuracy for all QoE related measures.
Biographie (en anglais)
Dr. Jérémie Leguay received a Ph.D. degree from Pierre & Marie Curie University where he worked jointly for the Computer Science LIP6 laboratory (Paris, France) and Thales Communications (Colombes, France) on ad hoc and delay tolerant networks. He was awarded the Thales Ph.D Prize for this work in 2007. From 2007 to 2014, he worked as a Research Engineer and as responsible for the Research & Technology lab on networked systems at Thales Communications & Security (Genneviliers, France) where he especially developed activities on sensor networks, mobile networks and software-defined networks. Jérémie Leguay has been the coordinator of the FP7 CALIPSO project on low-power protocols for the Internet of Things and the FP7 iTETRIS project on vehicular networks. In 2014, he joined the Mathematical and Algorithmic Sciences Lab at Huawei Technologies (Boulogne-Billancourt, France) at Principal Engineer and leader of the Network and Traffic Optimization team. His current research activities are on the control and management of IP and optical networks using optimization and machine learning tools.