采用Kohonen神经网络的ATM接入控制研究

Call Admission Control by Kohonen Neural Network in ATM Network

提出了一种新的采用Ｋｏｈｏｎｅｎ神经网络的ＡＴＭ接入控制方法，它避免了在ＢＰ神经网络中存在的大量训练数据和精确的学习训练之间的矛盾，不仅学习收敛快，而且数值结果表明在保证业务质量的前提下获得了相当高的信道利用率。

In this paper, we propose a new method for the call admission control in ATM network on the basis of the Kohonen neural network.It avoids the contradiction between accurate training and a large number of training data in BP neural network. It not only has a fast learning convergence rate,but also provides a high utilizing rate of channel and assures the quality of services in.the simulation.