用神经网络预测负荷的路由选择方法

A Routing Approach Based on Load Prediction Using Neural Network

电信网路由选择方法的优劣直接影响着网络的接通率和负荷平衡程度 .我国电信网的接通率只有45 %左右 .据介绍 ,若其接通率提高一个百分点 ,收益可达 10亿元 .本文针对目前所使用的路由选择方法的不足 ,提出基于神经网络预测的新的路由选择方法 ,包括性能指标、选路思想和递归神经网络预测等 .然后 ,分析和比较仿真结果 .这个方法因良好的分布特性和智能决策能力而优于其它方法 。

Routing approach influences the switch rate and load equilibrium of network directly.In China mainland,network switch rate is about 45%.It is said that raising one percent of switch rate of current telecommunications network will result in revenue almost one billion Yuan RMB.In the paper,on the bases of demerit-analysis of route-selecting methods being used in switch,a new routing approach based on neural network prediction was presented,including performance indicators,routing method,and recurrent neural network prediction.After that,simulation results were analyzed and compared.It′s better by virtue of its better distribution and intelligent decision ability,and provides excellent solution to improve network switch rate and balance network load.