准自相似业务流的SPN模型及其评价

Modeling and Evaluation of Pseudo Self-Similar Traffic with Stochastic Petri Nets

该文研究了近似自相似业务的马尔可夫模型,提出了该模型的相位型表达式并据此将模型从离散时间域转换到连续时间域,笔者对一个处理自相似业务到达的简单队列系统进行了研究,比较了系统在实际跟踪的仿真情况下和模型数值分析情况下的性能,模型数值分析用无穷状态随机Petri网实现,研究表明:马尔可夫模型虽能较好地近似自相似性,但仍需进行改进。

This paper addresses the suitability of a recently suggested approach for approximating self-similar traffic with a Markovian model.The phase-type nature of the proposed approach is identified and used to transform it from the discrete-time to the continuous-time domain.This paper then investigates the performance of a simple queuing system subject to self-similar arrival traffic,thereby comparing the result of trace-driven simulation with a measured self-simi-lar trace to those derived from a numerical analysis of suggested model.The numerical investigations are performed us-ing a special class of stochastic Petri nets which is particularly suitable for analyzing queuing-model like situations.The result indicates that the suggested Markovian traffic model needs still to be improved,even though the properties of self-similar are well approximated.