基于神经网络的动态路由选择算法  

A Dynamic Routing Algorithm Based on Neural Network

作　　者：袁友伟[1] 余江鸿[1] 湛含辉[1] 

机构地区：[1]株洲工学院计算机系,株洲412008

出　　处：《计算机工程与应用》2002年第12期165-167,共3页Computer Engineering and Applications

基　　金：湖南省自然科学基金资助项目(编号:00JJY2082)

摘　　要：在分析了网络中基于QoS组播路由问题的基础上,文章给出了基于Hopfield神经网络的动态路由选择算法的模型。仿真研究表明该算法具有良好的分布特性和智能决策能力,此方案不仅保证了带宽、端到端延时和延时抖动,优化了路由树的代价,而且有效地控制了算法的复杂性,是一种快速动态组播路由算法,能实现全局网络资源利用的优化,容易扩展到大型网络中应用。On the basis of describing the optimization of the multicast routing with QoS constrained,the paper presents a new approach based on the Hopfield model of artificial neural networks to solve the routing problem in the distribut-ed computer networks design.The simulation result shows that the routing algorithm is better by virtue of its better dis-tribution and intelligent decision ability.The scheme not only guarantees the bandwidth,end-to-end delay and delay-jin-ter,optimizing the cost of the routing tree,but also efficiently controls the algorithm complexity for large size of network.As a result,the proposed neural-network approach is suitable to be integrated into overall topological design processes,for moderate and high-speed networks subject to quality of services constraints as well as to changes in configuration and link costs.

关 键 词：神经网络 动态路由选择算法 服务质量 INTETNET 组播 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]