Minimização do conjunto de observadores para geração de matrizes de tráfego

• Ano de defesa: 2017
• Autor(a) principal: Silveira, Fernando Ávila Fossi
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Informática; Centro Tecnológico; UFES; Programa de Pós-Graduação em Informática
• Palavras-chave em Português: Traffic matrices; Data streaming algorithms; Combinatorial optimization; Matrizes de trafego (Computação); Algoritmos de processamento de cadeia de dados; Algoritmos; Otimização combinatória; Redes de computadores - Gerência; Ciência da Computação; 004
• Link de acesso: http://repositorio.ufes.br/handle/10/9853

Actual society rely on computer networks, so the management of these networks is an essential task. In order to manage a computer network it is necessary to know how information travels through it. A way to obtain this knowledge is by generating traffic matrices, which indicates the traffic volume exchanged between each pair of devices in the network. To generate these matrices is necessary to install observers in order to measure the traffic in the links, however this is a very costly operation. In recent years, data streaming algorithms based on probabilistic data structures have enabled traffic monitoring at a low computational cost. The manipulation of probabilistic data structures in observers can be done quickly and with little memory usage. However, even with probabilistic data structures, it is still necessary to install observers on all the devices to be monitored in the network in order to generate the traffic matrices, which leads to high implementation costs. In this context, this master’s thesis proposes and defines the problem of Minimizing the Set of Observers for Generating Traffic Matrices (MCO-MT), which consists of minimizing the number of observers installed in the network, that is, minimizing the size of the set of observers required in generating traffic matrices using probabilistic data structures. The problem is modeled and solved as a Minimum Set Covering Problem. In addition to the definition of MCO-MT, this work proposes: i) the development of a tool, called BitMatrix, for simulating and validating the generation of traffic matrices using data streaming algorithms based on probabilistic data structures; ii) two algorithms to solve the MCO-MT: a greedy heuristic and a greedy randomized heuristic combined with a local search to create a Greedy Randominzed Adpatative Search Procedure (GRASP), and iii) an data streaming algorithm to generate traffic matrices from a reduced set of observers. Extensive computational experiments are presented to validate the effectiveness of the proposed solutions.