RED算法的随机模型和仿真  

Stochastic Modeling and Simulations of the RED Algorithm

作　　者：罗超[1] 陈宗海[1] 薛福珍[1] 

机构地区：[1]中国科学技术大学自动化系,安徽合肥230027

出　　处：《中国科学技术大学学报》2005年第6期947-953,共7页JUSTC

摘　　要：运用随机分析的方法建立了一个关于RED(randomearly detection)队列长度及其指数加权滑动平均值(exponentially weighted moving average ,EWMA)的随机模型.对该模型的动态仿真和数值分析表明,它能够很好地预测RED在忙期的动态行为.利用这个模型,RED的参数可以通过简单的数值分析方法进行设置,避免了经验方法在链路特性参数和流量参数变化时所出现的性能问题.A stochastic model of the random its exponentially weighted moving average early detection（RED） queue length and （EWMA） value is developed based on stochastic analysis. By comparing research between dynamic simulation and numerical analysis, it has been verified that this model can predict the busy-period dynamics of RED very well, which makes it possible to set up RED parameters by simple numerical analysis, thus avoiding the performance problems that occur when links and traffic parameters oscillate and empirical methods are used.

关 键 词：RED 拥塞控制 EWMA 队列长度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]