Massive Interactive Multimedia Services over the Internet
Résumé
The Internet is now a support infrastructure for the entertainment industry. In particular, the demand for interactive, multimedia applications over the Internet has significantly grown over the past ten years, i.e. since I defended my PhD thesis. As a researcher in the field of multimedia and Internet, I found the last decade especially exciting and challenging. My contributions to a better understanding and, hopefully, a better implementation of large-scale, interactive, multimedia services have taken various forms, which I classify here into three categories: measurements, optimization, and system design. The manuscript describes two representative studies in each of these categories. As for the measurement chapter, I present two campaigns of measurement related to gaming. In the first one, the motivation is to study the latency of cloud gaming applications, where the game engine is hosted in a datacenter and the output streamed to the end-users. In the second one, a popular platform allowing gamers to broadcast themselves playing (live gameplay), namely Twitch.tv, is dissected. The optimization chapter deals with live streaming delivery. The first study aims at minimizing the waste of resources for systems based on multiple concurrent peer-to-peer overlays, while the second study addresses the reduction of cost for Content Delivery Networks having to deliver rate-adaptive, live, video streams. The topic of the system design chapter is related to Content Centric Networks, where the main functions of the networks are based on the content name instead of the content location. I present here our proposal to augment the CCN protocol in order to better leverage cache storage inside routers. The second study describes an evaluation of performances of this protocol based on experiments in laboratory. The manuscript ends with a chapter about research perspectives in the field. Although I doubt that it is possible to have a clear vision of what will the research look like in a horizon of ten years, especially in my field, I present some global research axis, which, in my opinion, are worth exploring in the next future years.
Origine : fileSource_public_fileSource
Loading...