Análisis del comportamiento autosimilar del tráfico Ethernet de las redes de datos

Abstract

The queuing analysis has been and is of enormous usefulness for the designers of networks and analysts of traffic, to effects of planning the capacities of the elements of network and of predicting his performance. These analyses depend on the nature Poisson of the traffic of data. Nevertheless, many results predicted from the analysis of queuings differ significantly from the performance observed in the reality. Diverse studies have demonstrated that for some environments the pattern of traffic is self-similar, instead of Poisson. This concept is related to other acquaintances since it are the fractals and the theory of the chaos. From beginning of the 90s were begun to publish documents referred to the self-similarity of the traffic of Ethernet. The present work develops successively the following paragraphs: 1) Self-similarity, 2) Self-similar data traffic, 3) Ethernet data traffic, 4) Case of experimental Study of Ethernet traffic, 5) Analysis with distribution Pareto, and 6) Analysis of goodness of fit with Kolmogorov-Smirnov's test.