Estimation of Hurst Index and Traffic Simulation

EasyChair Preprint 6343

Abstract

The investigation of traffic properties of modern networks requires new approaches, the use of adequate types of distributions of traffic components, and measurement errors should be also taken into account. The models of the request flow are approximated by different distributions with "light tails" (Gaussian, Poisson distributions) as well as "heavy tails" (Pareto, Weibull, log-normal distributions). Self-similar traffic models are widely used to describe traffic in packet-switched networks. The degree of self-similarity of traffic can be determined by various methods, one of them is the estimation of the Hurst index. In the paper, new approaches in simulation of self-similar traffic and theoretical estimation of Hurst index with measurement errors are studied, the statistical simulation of main needed distributions with heavy tails is also considered.

Keyphrases: Hurst index, accuracy and reliability of simulation., fractional Brownian motion, self-similar traffic

@booklet{EasyChair:6343,
  author    = {Anatolii Pashko and Iryna Rozora and Olga Sinyavska},
  title     = {Estimation of Hurst Index and Traffic Simulation},
  howpublished = {EasyChair Preprint 6343},
  year      = {EasyChair, 2021}}