闭排队网络当性能函数与参数相关时的性能灵敏度分析

Sensitivity Analysis of Performance with Parameter-Dependent Performance Functions in Closed Queueing Networks

借助于无穷小矩阵摄动方法 ,讨论了一类Markov过程 ,其稳态性能关于参数摄动的灵敏度分析问题 .然后研究了闭排队网络的稳态性能灵敏度分析问题 ,并在参数相关性能函数的情况下 ,给出了网络的几种稳态性能的灵敏度公式 .这些公式表明稳态性能灵敏度很容易通过网络势能进行计算 .

By using the approach of the infinitesimal generator perturbation, the problems of sensitivity analysis of the steady_state performance with respect to the parameter perturbation are discussed for a class of Markov processes. Then the sensitivities of steady_state performance are studied in closed state_dependent queueing networks. Sensitivity formulas of several kinds of the steady_state performance are given for parameter_dependent performance functions. These formulas show that the sensitivities of steady_state performance can be easily calculated by the potentials of networks.