宽带网中基于正则化模糊神经网络的连接接纳控制

Connection Admission Control in Broadband NetworksBased on Normalized Fuzzy Neural Networks

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()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CAC力。通过合理地选择输入语言变量和设计模糊规则学习结构,以使对连接的接受拒绝作出正确决定,并保证服务质量()。仿真结CAC/QoS果表明,与现有的各种相比,该方法可获得更高的资源利用率、更大的吞吐量和更低的信元丢失率。

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.