摘要
计算机网络具有既随机又相关的特点，因此选择计算机网络性能评价的数学工具要考虑到网络数据流的特性．排队模型是一种确定性的数学模型，用确定性的模型描述网络既随机又相关的动态行为是不够的．ＰｅｔｒｉＮｅｔｓ（ＰＮ）能够比较方便地刻画网络的相关事件，描述网络的竞争、碰撞和阻塞，从ＰＮ进行扩充发展而来的随机ＰＮ，可以比较方便的刻画网络事件的随机性，将二者结合起来从相当程度上缓和了计算机网络建模与分析之间的矛盾，是一种颇具吸引力的建模工具．
Computer networks have random and correlative characteristics. The choice of the mathematical tools for the performance evaluation of computer networks should consider the characteristics of data flow. This paper compares some mathematical tools for this challenge. Queueing model, which is often used to analyze the performance of computer systems, has its limit for describing the events which are characterized as both random and correlative, especially analyzing the end-to-end performance if the condition of product form solution is not satisfied. Petri Net is appropriate to characterize the correlative events of networks. Introducing the stochastic process into Petri Nets can describe the random characteristics of network events, and ease the contradiction between modeling and solution.Thus, Stochastic Petri Nets is an attractive modeling tools for the performance evaluation of computer networks.
出处
《计算机学报》
EI
CSCD
北大核心
1996年第6期409-420,共12页
Chinese Journal of Computers
关键词
性能评价
计算机网络
随机过程
建模
Performance evaluation
computer networks
stochastic process
Markov process
queueing model
Petri Nets
Stochastic Petri Nets.
