论文部分内容阅读
研究了用马尔可夫调制的泊松过程(MMPP)对Internet多分形流量突发行为进行近似建模的能力.MMPP可用于描述适当时间尺度范围内流量的变化以及相关性,而且它可作为排队系统输入过程得到分析结果.描述了刻画突发流量行为的重要统计量,在此基础上给出了一个基于矩的MMPP参数估计方法.除了对MMPP进行拟合优度检测以外,本文通过将MMPP的样本过程和实际流量记录输入到排队系统模型中比较其输出结果采研究MMPP对排队性能的预测能力.数值和仿真实验表明,MMPP能够较好地用于对多分形流量近似建模,即可以准确地预测网络结点的排队性能.
This paper studies the ability of approximating Internet multi-fractal traffic burst behavior with Markov-modulated Poisson process (MMPP), which can be used to describe the change of traffic and its correlation within the appropriate time scale, and it can be used as queuing The system input process is analyzed.The important statistic to describe the behavior of the burst traffic is described.On the basis of this, a moment-based MMPP parameter estimation method is given.In addition to the goodness of fit test for the MMPP, Of the sample process and the actual flow record into the queuing system model to compare the output of its output to study the predictive ability of MMPP queuing performance.Numerical and simulation experiments show that the MMPP can be used to approximate the multifractal flow modeling, Accurately predict the queuing performance of network nodes.