Transmissões de vídeo ao vivo geradas por usuários: caracterização e análise

• Ano de defesa: 2009
• Autor(a) principal: Thiago Henrique Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Processamento de imagens v; Sistemas de transmissão de dados; Computação; Cliente/servidor (Computação); Processamento de imagens Técnicas digitais
• Link de acesso: http://hdl.handle.net/1843/SLSS-7WMGAQ

In the past few years, we have observed a huge revolution in terms of the way we create and distribute video on the Internet. With the popularization of digital cameras and with the increase of Internet access speeds, video content creation and distribution on the Internet has grown considerably. There are a number of recently created companies which offer the users the possibility of transmitting, free of charge, live streaming video using Web 2.0 based systems. In spite of their short operation time on the Internet, it is already possible to observe a significant popularity increase of those services. In this context, this dissertation aims to propose and evaluate solutions that contribute to improve the scalability of this type of system. This work encompasses two major steps. First, we collected data from a popular online live video sharing service, and provided a characterization of key aspects of user dynamic behavioral patterns. We determine statistical models for, among other aspects, session durations, new session arrival process during a transmission, and duration of transmissions performed by a channel.Next, our characterization findings were used to drive the design of a simulator of different content distribution architectures, including the traditional client-server and Peer-to-Peer (here referred to as pure P2P) solutions. Our characterization findings also motivated the proposal of two new hybrid strategies, which combine the benefits from those two architectures. We then performed an extensive evaluation of the four strategies, considering different scenarios as well as metrics that reflect both the interests of the users (e.g., continuity index and startup delay) and of the server (network bandwidth). Our results indicate that the hybrid strategies yield the best tradeoffs between quality of service and resource consumption.