Design, implementation and verification of the QoES platform for QoS configuration based on qoe feedback

Resumen

The Internet is becoming more and more accessible to a wide spectrum of devices. Nowadays users demand a huge quantity of media contents: video streaming at High Definition (HD) quality, videoconferencing, high interactive games, web pages with dynamic content, embedded video players… The variety of Internet services offered to the end-user keeps on growing, as anyone can access them from any Internet access point. In addition, the bandwidth resources that these services need are increasing as well. No matter which the contracted bandwidth between household and Internet Service Provider (ISP) is: when it is simultaneously shared by several services there is a link congestion, more pronounced when the contracted bandwidth is not very high (e.g. ADSL links). Consequently, a Quality of Service (QoS) management is necessary in order to distribute the available link resources between consumed services at the shared link. Moreover, the QoS management has to take into account the performance of each service for a set of QoS parameters. In this context, since services are consumed by users, it is also important to measure if a good user Quality of Experience (QoE) is being achieved. Note that although the measured parameter is the experience, different factors affecting it can be highlighted. QoE takes into account both the subjective factors (e.g.: user perception, experience and expectations) and objective parameters such as application and QoS parameters. This combination causes the same set of QoS values to be mapped to different QoE on different users and vice versa. Thus, effective QoS management must be based on the QoE. This work details the design, implementation and verification of QoES (Quality of Experience and Service) platform. The intention of QoES is to provide a QoS configuration at an access node based on QoE user feedback. In order to reach this purpose, this work starts studying the current state of the art. This includes the QoS / QoE relationship, the standards and technical recommendations, measurement methods reported by standardization institutions and literature, and the related work on the adaptation of multimedia services to QoS fluctuations. Once the analysis of the state of the art is done, a procedure (including the metric and algorithms) to evaluate the QoE is proposed, which has been implemented and integrated in the QoES platform. Moreover, as a previous service traffic pattern characterization is also important for the definition of QoE tests, service analysis tools for the network layer have also been included as part of the QoES platform. For the cases where application layer parameters are also of interest, the capability of including ad-hoc software as a plug-in (add-on) to the QoES platform has also been incorporated. In particular, the add-on for analysing YouTube service has been implemented as a use case, and its adaptive traffic pattern has been studied. Finally, this thesis shows the verification process by including QoE tests for Onlive cloud gaming and Yomvi streaming services, where the correspondence between the QoE outcomes and the monitored network parameters is discussed. Based on the QoE feedback received, the final QoS configuration is also calculated and applied at HG. This thesis work finishes with the discussion of the main conclusions and the identification of the future lines.