Português English
Contato

Tese de Roberto Irajá Tavares da Costa Filho


Detalhes do Evento


Aluno: Roberto Irajá Tavares da Costa Filho
Orientador: Prof. Dr. Luciano Paschoal Gaspary

Título: Scalable Path Selection Based on QoE Prediction in Software-Defined Mobile Networks
Linha de Pesquisa: Redes de Computadores

Data: 05/04/2019
Horário: 09h
Local: Sala 218 do Prédio 43412 do Instituto de Informática da UFRGS

Banca Examinadora:
– Prof. Dr. Alberto Egon Schaeffer Filho (UFRGS)
– Prof. Dr. Eduardo Coelho Cerqueira (UFPA)
– Prof. Dr. José Ferreira de Rezende (UFRJ)

Presidente da Banca: Prof. Dr. Luciano Paschoal Gaspary

 

Abstract: Ranging from traditional video streaming to Virtual Reality (VR) videos, the demand for video applications to mobile devices is booming. To deal with the massive traffic produced by video applications, mobile operators rely on offloading technologies such as Small Cells, Content Delivery Networks and, shortly, Cloud Edge and 5G Device to Device communications. Although these techniques are fundamental for improving network efficiency, they produce a multitude of paths through which the user traffic can be forwarded. Most importantly, such an increased path diversity does not provide any guarantee regarding user’s Quality of Experience (QoE). Thus, a critical problem arises about how to handle the increasing video traffic while managing the interplay between infrastructure optimization and QoE. Solving this issue is remarkably difficult, and recent investigations do not consider the large-scale context of mobile operator networks. In a nutshell, the problem of dynamic provisioning of QoE-aware paths can be decomposed into two fundamental functions: (i) QoE measurement or estimation and (ii) path selection on a programmable network. To address the problem of QoE estimation, we propose a model to predict video streaming quality based on the observation of performance indicators of the underlying IP network. To accomplish this objective, the proposed model leverages lightweight active measurements and machine learning techniques. In a further step, we introduce a novel QoE-aware path deployment heuristic for large-scale SDN-based mobile networks. The scheme relies on both a polynomial-time algorithm for composing multiple QoS metrics and a scalable QoS to QoE translation strategy. Obtained results show that the proposed methods for video streaming performance prediction produce accurate estimates. As a consequence, our approach for QoE-aware path selection outperformed state-of-the-art techniques approaches by reducing impaired videos in aggregate QoE by at least 37% and lowering accumulated video stall length four times. Based on the lessons learned with QoE prediction for traditional video streaming, we finally explore the VR video domain by introducing PERCEIVE and VR-EXP. PERCEIVE is a two-stage method for predicting the perceived quality of adaptive VR videos when streamed through mobile networks. By means of machine learning techniques, our approach is able to predict the playout performance of adaptive VR video and use this information to model and estimate QoE. In turn, VR-EXP consists of an experimentation platform that allows in-depth evaluation of state-of-the-art VR video optimization techniques. VR-EXP relies on software-based emulation to assess the interplay between a set of VR video optimization techniques and different levels of network performance.

Keywords: Quality of Service. Quality of Experience. Video Streaming. Virtual Reality. VR Video. Path Selection. Mobile Networks.