摘要随着计算机网络的飞速发展，单纯的端到端的拥塞控制已经难以确保网络的高效运行。网络的中间设备，如路由器也要参与拥塞控制的工作，对队列进行管理。近年来，作为一种有效的拥塞控制方法，主动队列管理机制(AQM)受到国内外广泛关注。AQM技术本质是对拥塞进行早期检测，使端系统在路由器缓冲区队列溢出之前降低发送速率，减少数据包的大量丢失。本文主要研究以下三个内容：7497
(1) 介绍RED算法的原理，利用NS2软件平台进行仿真实验，验证RED算法的拥塞控制性能。
(2) 针对RED算法的缺陷，介绍改进的自适应算法ARED，进行仿真实验，并与RED算法进行比较。
(3) 提出一种基于单神经元的自适应AQM算法，并从队列稳定和丢包率等性能指标对该算法和RED算法进行仿真比较。
关键词 拥塞控制 主动队列管理 RED算法 神经网络
毕业设计说明书（论文）外文摘要
Title Neural-Network-Based Adaptive RED Algorithm and Simulation
Abstract
With the rapid development of computer network, simple end-to-end congestion control has been difficult to ensure efficient network operation. So the middle equipment of the network such as routers also participates in congestion control with queue management. As an effective congestion control method, Active Queue Management (AQM) has absorbed so much attention from home and abroad in recent years. AQM technology is essentially the early detection of congestion, which makes the client system reduce the sending rate before the router buffer queue overflow to reduce the loss of a large number of packets. This paper mainly studies the following three elements:
First, this paper will introduce the principle of RED algorithm and use the NS2 software for simulation to verify the congestion control performance of RED algorithm.
Second, as to the defects of RED algorithm, the paper will introduce an improved adaptive RED algorithm (ARED) and do some simulation experiments to compare it with RED algorithm.
Third, the paper will propose a single-neuron-based adaptive AQM algorithm. Then the paper will do some simulation experiments to compare the queue stability and packets loss rate of the proposed algorithm and RED algorithm.
Keywords congestion control AQM RED neural network
目录
1 绪论 1
1.1 网络拥塞控制的研究现状 1
1.2 主动队列管理算法的发展及分类 4
1.3 本文主要内容安排 9
2 随机早期检测算法及其仿真研究 11
2.1 RED算法的设计原理 11
2.2 网络模拟软件NS介绍 13
2.3 RED算法的仿真研究 15
2.4 本章小结 19
3 自适应RED算法及其仿真研究 21
3.1 ARED算法的设计原理 21
3.2 ARED算法的仿真研究 22
3.3 本章小结 26
4 基于单神经元的AQM算法及其仿真研究 27
4.1 基于单神经元的TCP/AQM系统模型 27
4.2 基于单神经元的AQM算法实现 29
4.3 稳定性分析 31
4.4 基于单神经元的AQM算法的仿真研究 34
4.5 本章小结 37 基于神经网络的自适应RED算法及其仿真研究:http://www.751com.cn/zidonghua/lunwen_5482.html