Performance analysis of multifractal network traffic
Performance analysis of multifractal network traffic
复制标题
多重分形网络流量的性能分析
DOI：
10.1002/ett.955
复制
发表时间：
2004
期刊：
影响因子：
--
通讯作者：
I. Maricza
中科院分区：
文献类型：
--
作者：
Trang Le Dinh Dang;S. Molnár;I. Maricza
本文在多重分形网络性能分析框架下，给出了一些新的结果，包括多重分形网络流量的表征和建模，以及多重分形网络的性能分析。首先，我们提出了一个新的多重分形网络流量模型的基础上的乘法级联与独立的对数正态过程相结合。这个流量模型能够提供一个非常准确的适合的多重分形特性的datatraffic，包括缩放函数和矩因子，但它也是足够简单，从实用的角度来看，只有三个参数。此外，该模型具有许多重要的性质，包括长期依赖（LRD），多重分形和对数正态性的数据集观察。我们还提出了一个近似的队列尾渐近在一个无限容量的单服务器队列服务在一个恒定的速度驱动的一般多重分形输入过程。我们表明，在单分形分数布朗运动（fBm）输入流量的特殊且重要的情况下，我们的结果给出了著名的韦伯尾巴。证明了具有标度性质的高斯过程类是单分形过程类，并给出了相应的特征函数。应用这种近似，我们为一般多重分形业务的性能估计提供了一种新的实用方法。最后，我们提出了一个实际的案例研究，以显示我们的框架测量数据流量的实际应用，并验证我们的多重分形模型和我们的测试结果。版权所有© 2004 AEI
In this paper we present some new results in the framework of multifractal performance analysis including both the characterization and modeling of multifractal network traffic, and also the multifractal queueing performance analysis. We first propose a new multifractal traffic model for network traffic based on the combination of a multiplicative cascade with an independent lognormal process. This traffic model is able to provide a very accurate fit to the multifractal characteristics of datatraffic including both the scaling function and the moment factor, but it is also simple enough from a practical point of view having only three parameters. In addition, the model features many important properties observed in datatraffic including long-range dependence (LRD), multifractality and lognormality. We also present an approximation for the queue tail asymptotics in an infinite capacity single server queue serviced at a constant rate driven by a general multifractal input process. We show that in the special and important case of the monofractal fractional Brownian motion (fBm) input traffic our result gives the well-known Weibullian tail. We prove that the class of Gaussian processes with scaling properties is in the class of monofractal processes and we derive the related characterization functions. Applying the approximation we provide a new practical method for queueing performance estimation of general multifractal traffic. Finally, we present a practical case study to show the practical application of our framework for measured data traffic and also to validate both our multifractal model and our queueing results. Copyright © 2004 AEI