Alocação adaptativa de banda e controle de fluxos de tráfego de redes utilizando sistemas Fuzzy e modelagem multifractal

• Ano de defesa: 2014
• Autor(a) principal: Cardoso, Alisson Assis
• Orientador(a): Vieira, Flávio Henrique Teles
• Banca de defesa: Vieira, Flávio Henrique Teles, Carvalho, Cedric Luiz de, Brito, Leonardo da Cunha
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Engenharia Elétrica e da Computação (EMC)
• Departamento: Escola de Engenharia Elétrica, Mecânica e de Computação - EMC (RG)
• País: Brasil
• Palavras-chave em Português: Análise multifractal; Modelagem Fuzzy; Predicão de tráfego de rede; Alocacão de banda; Controle de tráfego de rede; Funções de base ortonormal; Orthonormal basis function
• Palavras-chave em Inglês: Multifractal analysis; Fuzzy modeling; Network traffic prediction; Band-width allocation; Traffic flow control
• Área do conhecimento CNPq: SISTEMAS DE COMPUTACAO::HARDWARE
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/3164

Inthispaperweproposeafuzzymodel,calledFuzzyLMScomAutocorrela¸c˜aoMultifractal, whose weights are updated according to information from multifractal traffic modeling. These weights are calculated by incorporating an analytical expression for the autocorrelation function of a multifractal model in the training algorithm of the fuzzy model that is based on the Wiener-Hopf filter. We evaluate the prediction performance of the proposed network traffic prediction algorithm with respect to other predictors. Further, we propose a bandwidth allocation scheme for network traffic based on the fuzzy prediction algorithm. Comparisons with other bandwidth allocation schemes in terms of byte loss rate, link utilization, buffer occupancy and average queue size verifies the efficiency of the proposed scheme. Also, We propose an other adaptive fuzzy algorithm, called Fuzzy-LMS-OBF com alfa adaptivo , for traffic flow control described by theβMWM model. The proposed algorithm uses Orthonormal Basis Functions (OBF) and its training based on the LMS algorithm. We also present an expression for the optimal traffic source rate derived from Fuzzy LMS. Then, we evaluate the performance of the Fuzzy-LMS-OBF com alfa adaptivo algorithm with respect to other methods. Through simulations, we show that the proposed control scheme is benefited from the superior performance of the proposed fuzzy algorithm. Comparisons with other methods in terms of mean and variance of the queue size in the buffer, Utilization rate of the link, Loss rate and Throughput are presented.