摘要
研究神经网络在ＡＴＭ流量预测中的应用，通过计数过程对多媒体业务流的比特率波动做出了准确预测。与传统方法相比较，该方法无须对网络进行复杂的数学分析，借助足够的学习样本，即可自动学习输入和输出之间的关系，预测网络业务量的将来值。文中实现了对单个业务及组合业务流量的准确预测，这对于ＡＴＭ流量管理具有十分重要的意义。
The application of neural network in the traffic prediction of ATM(asynchronous transfer mode)networks is explored in this paper.An accurate prediction of the multimedia traffic via its count process is given.Compared with the traditional methods,this new way does not need any complicated mathematical calculations,and it can automatically learn the relationship between input and output patterns, and predict the future traffic values. Accurate traffic predictions both of the single source and the mixture source have been achieved,done in this paper, which is of great importance to the traffic management of ATM.
出处
《桂林电子工业学院学报》
1998年第4期11-14,共4页
Journal of Guilin Institute of Electronic Technology
基金
广西青年科学基金
关键词
异步转移模式
ATM
神经网络
流量预测
计数过程
asynchronous transfer mode (ATM),neural network, traffic prediction, count process
