论文部分内容阅读
近年来,ATM网络中路由发现和拥塞控制的资源优化分配方法正在得到广泛的关注。本文提供的解决方案是基于神经网络策略的,文中描述的方案采用多变量约束优化算法来处理ATM网络中UNI/NNI(用户网络接口/网络网络接口)的业务需求,从而找到近似最佳的PVC/SVC(永久虚电路/交换虚电路)路由。 自动回归反向传播网络的预知流量规定、用户定义的业务参数以及网络负荷条件等都限制了优化的程度。利用计算机仿真神经网络,使其在评估网络性能方面进行训练和回忆,这样,使系统按照所期望的目标运转,逐渐达到预期网络性能指标。
In recent years, resource allocation and optimization methods for route discovery and congestion control in ATM networks are gaining widespread attention. The solution provided in this paper is based on the neural network strategy. The solution described in this paper uses the multivariate constrained optimization algorithm to deal with the business requirements of UNI / NNI (User Network Interface / Network Network Interface) in ATM network, so as to find the approximate best PVC / SVC (permanent virtual circuit / switched virtual circuit) routing. Predictive traffic regulations that automatically return to backpropagation networks, user-defined business parameters, and network load conditions all limit the degree of optimization. The use of computer simulation of neural networks, to make it in the assessment of network performance training and memory, so that the system in accordance with the desired goal of operation, and gradually reach the expected network performance indicators.