一种基于ANFIS的ABR拥塞控制

Congestion Control Method of ABR Traffic Based on ANFIS

由于ATM网络环境的复杂性、多变性,用常规的数学模型对网络模型、可用带宽的获取以及控制器设计的描述具有很大的局限性,因此论文提出了一种基于自适应模糊推理系统(AdaptiveNeuralFuzzyInferenceSystems,ANFIS)的ABR业务拥塞控制方法,该方法结合模糊推理系统的规则结构化及神经网络强泛化能力的优点,克服了模糊推理模型的偶然性和神经网络收敛速度慢、训练时间过程长等缺点。仿真结果表明使用ANFIS进行拥塞控制的可行性,增加了系统稳定性并减小了信元丢失率。

Due to the complexity and variability of ATM,It is difficult to describe the congestion model,bandwidth acquiring and congestion controller designing under traditional math model.In this paper,a method based on ANFIS （Adaptive Neural Fuzzy Inference Systems） is presented for ABR service congestion control,which takes the both advantage of the regular rules of fuzzy inference systems and the adaptation of BP neural network,overcome the indetermination of system model using fuzzy theory and the slow convergence of training using neural network.The simulation experiments show that congestion control based on ANFIS is very effective to improve the stability and the low cells lose rate of system.