动态无级网状态概率的卷积算法

Convolution Algorithm for the State Probabilities of Network with Dynamic Non-Hierarchical Routing

通过对集中控制、状态相关控制方式的动态无级交换网进行模型化,提出了一种求解电路群状态概率的精确算法——卷积算法.算法的关键是一个全局状态相关到达转换为局部状态依存输入过程的公式.本方法的计算结果不仅可以方便地求得网中任意端对间各种阻塞概率以及电路群的效率等性能分析参数,而且还可以仅对所关心的端对进行局部评价,从而避免了由状态转移概率矩阵导出系统稳态概率的复杂计算.

By modeling the DNHR network with state-dependent control in centralized way, this paper advances a convolution algorithm for the exact calculation of the state probabilities of circuit groups.The essentiality of the algorithm is a set of formulae named correlated ar- rival-local state-dependent input transfomation.The resulting model would make it easy and convenient for us to evaluate parts of a network which may be important for our study and to calculate various performance analytical parameters such as end to end congestions.