基于前馈神经网络的交换网络路径选择控制器  

Feedforward Neural Network Based Shortest Path Controller for Multi Stage Switch Networks

作　　者：纪晓东[1] 周继成[1] 王德隽[1] 

机构地区：[1]北京邮电大学培训中心

出　　处：《北京邮电大学学报》1996年第3期89-92,共4页Journal of Beijing University of Posts and Telecommunications

摘　　要：首先提出了一种基于前馈网络的多模拟输入最小值选择电路，然后将该网络用于解决多级级联交换网络的最小费用路径选择问题．计算机仿真表明，此前馈神经网络总能找到全局最优．由于前馈神经网络的计算时延非常短。A feedforward neural network was first proposd, which can select the minimun among several analog inputs. Then a shortest path controller, based on this neural network, for multi stage switch networks, was develope. Computer simulation results show that the neural networks can find the global minimum in both cases.

关 键 词：神经网络 级联交换网络 路径选择 通信网 

分 类 号：TN913.11[电子电信—通信与信息系统] TN913.2[电子电信—信息与通信工程]