Fuzzy time series aplicada na predição de QOS em redes de computadores

• Ano de defesa: 2018
• Autor(a) principal: Francirley Resendes Borges Costa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Engenharia elétrica; Predição (Lógica); Redes de computadores
• Link de acesso: http://hdl.handle.net/1843/BUBD-AYJJES

Increasingly, there is a need for quality assessment tools in a number of areas. Just as in computer networks, mainly due to the large increase in consumption of services over the internet. In this context, this work presents an approach with Fuzzy Time Series to predict quality of service. For this purpose a new method is proposed inspired by the pioneering methods of Fuzzy Time Series. The characteristics of computer networks are studied and mechanisms, such as linear regression, are applied to the techniques of the literature methods in an attempt to improve the accuracy of the approach. Statistical tests were designed, based on data collected in real environments, in order to validate the proposal. The results are encouraging, especially considering the trend change rate of the proposed method, which reached 87.16%, thus indicating that it is possible to use an approach of a quality of service forecast in computer networks together with fuzzy time series.