宽带网中基于正则化模糊神经网络的连接接纳控制

提出了一种宽带网络中基于模糊神经网络的连接接纳控制（CAC）方法，结合了模糊逻辑的语言控制能力和神经网络的自学习能力，通过合理地选择输入语言变量和设计模糊规则学习结果，以使CAC对连接的接受／拒绝作出正确决定，并保证服务质量（QoS）。仿真结果表明，与现有的各种CAC相比，该方法可获得更高的资料利用率，更大的吞叶量和更低的信元丢失率。

Connection Admission Control in Broadband NetworksBased on Normalized Fuzzy Neural Networks

This paper proposes a fuzzy neural network approach for connection admission control (FNN-CAC) in broadband networks. With reasonable input language variables and rule structure of the FNN-CAC, it can make accurate decision of accept/reject for the new connection with guaranteed QoS. Simulation results show that the proposed FNN-CAC can produce small average queue length and superior link utilization while maintaining a high throughput.