Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-06-06T14:59:44Z No. of bitstreams: 1 biancaportesdecastro.pdf: 1203353 bytes, checksum: 0cd5843bff9e747e5432fff99ec1e565 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-06-07T11:04:29Z (GMT) No. of bitstreams: 1 biancaportesdecastro.pdf: 1203353 bytes, checksum: 0cd5843bff9e747e5432fff99ec1e565 (MD5) / Made available in DSpace on 2017-06-07T11:04:29Z (GMT). No. of bitstreams: 1 biancaportesdecastro.pdf: 1203353 bytes, checksum: 0cd5843bff9e747e5432fff99ec1e565 (MD5) Previous issue date: 2014-08-25 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Aplicações multimídia são muito populares na internet. Grande parte delas necessita de multicast para escalar. É sabido que multicast em nível de redes não foi implementado como desejado. Protocolos em nível de aplicação são a solução atual. Apesar do sucesso dos protocolos ALM (Application Layer Multicast), a maioria dos protocolos existentes são custosos e acarretam grande sobrecarga de controle à rede. Neste trabalho, apresentamos um novo protocolo de fluxo contínuo baseado em árvore, utilizando a desigualdade triangular entre cada três peers para gerenciamento dinâmico da topologia (o PALMS+). O novo protocolo é simples e com baixa sobrecarga. Mesmo assim, seu desempenho é tão bom quanto o estado da arte. Experimentos realizados na plataforma Oversim (OMNet++) demonstraram que o PALMS+ manteve desempenho tão bom quanto o estado da arte (e.g. protocolo NICE), mesmo quando submetido a alto churn em uma rede heterogênea. De fato, a sobrecarga nos peers do novo protocolo é menor que 10% da sobrecarga gerada pelo NICE. O protocolo PALMS+ entrega os dados em menos de 1,5s. O novo protocolo mostra-se adequado a vídeo ao vivo, escalando mesmo em cenários realistas e com alto churn. / Multimedia applications are very popular on the internet. Many of these applications need multicast to scale. However, network layer multicast has not been implemented in the internet. Application layer multicast (ALM) protocols are a practical alternative. However, despite their popularity, many existing ALM protocols and mechanisms are expensive and bring a large overhead control on the network. In the present work, a new protocol is proposed for content distribution based on tree, using the triangular inequality between every three peers to dynamic topology control (the PALMS+). The new protocol is simple and with low overhead. Nevertheless, its performance as good as the state of the art. Experimental results conducted with the OverSim platform (OMNet++) suggest that PALMS+ improves the performance of a state-of-art implementation of ALM protocol when compared against the NICE protocol. Furthermore, the control message overhead at peers using the PALMS+ protocol is reduced by 10%, when compared with NICE. In the PALMS+ protocol, chunks are delivered up to 1,5s. Results confirm that proposed implementation of PALMS+ is very suitable to real-time video streaming, even when churn is high.

No description available.

This work focuses on statistical methods to understand how frequently rare events occur and what the magnitude of extreme values such as large losses is. It lies in a field called extreme value analysis whose scope is to provide support for scientific decision making when extreme observations are of particular importance such as in environmental applications, insurance and finance. In the univariate case, I propose new techniques to model tails of discrete distributions and illustrate them in an application on word frequency and multiple birth data. Suitably rescaled, the limiting tails of some discrete distributions are shown to converge to a discrete generalized Pareto distribution and generalized Zipf distribution respectively. In the multivariate high-dimensional case, I suggest modeling tail dependence between random variables by a graph such that its nodes correspond to the variables and shocks propagate through the edges. Relying on the ideas of graphical models, I prove that if the variables satisfy a new notion called asymptotic conditional independence, then the density of the joint distribution can be simplified and expressed in terms of lower dimensional functions. This generalizes the Hammersley- Clifford theorem and enables us to infer tail distributions from observations in reduced dimension. As an illustration, extreme river flows are modeled by a tree graphical model whose structure appears to recover almost exactly the actual river network. A fundamental concept when studying limiting tail distributions is regular variation. I propose a new notion in the multivariate case called one-component regular variation, of which Karamata's and the representation theorem, two important results in the univariate case, are generalizations. Eventually, I turn my attention to website visit data and fit a censored copula Gaussian graphical model allowing the visualization of users' behavior by a graph.