基于随机早期检测算法的网络拥塞控制系统的分岔及混沌

基于传输控制协议和随机早期检测拥塞控制算法，提出多路由器网络拓扑下系统的改进型离散时间映射，采用解析及数值方法研究该映射中出现的分岔现象及混沌，给出系统的稳定性条件．借助分岔图讨论稳定区域附近参数变化对系统特性的影响，给出边界碰撞分岔产生的条件．研究结果表明，系统参数的微小变化会导致系统动力学特性的频繁改变：边界碰撞分岔导致队列长度的大幅度振荡，在增大端到端的时延抖动、降低链路利用率的同时，使得系统对参数表现出极端敏感性；在配置系统参数时，应尽量避免亚临界分岔的发生．

Bifurcations and Chaos in Internet Congestion Control System Based on Random Early Detection

A discrete model for a simplified multi-router transfer control protocol network with random early detection congestion control system is proposed by modifying the previously developed one, where the nonlinear behaviors occurring in the system are researched by analytical and numerical methods, and the stability of the fixed points in different parameters is obtained simultaneously. According to the bifurcation diagram, the effects of parameters disturbance on system performance are discussed, and the sufficient condition of occurring border collision bifurcation is given. It is shown that the nonlinear characteristics change frequently with the slight variations of parameters, and the border collision results in remarkable oscillations of queue length and the parameter sensitivity to increase the end-to-end delay and reduce the link utilities, thus it is necessa- ry to avoid sub-critical bifurcation as designing system parameters.