一种基于Smith预估器的主动队列管理(AQM)拥塞控制算法

An AQM congestion control algorithm based on Smith predictor

有效的拥塞控制机制是保证Internet稳定运行的关键因素之一,网络拥塞控制系统本质上是一个时滞系统,传输时延是网络拥塞控制必须考虑的一个重要因素.本文应用Smith预估控制原理,在进行适当模型拟合处理的基础上,提出了一种基于Smith预估器的主动队列管理(AQM)算法(AQMAlgorithmbasedonsmithpredictor算法,简称Smith-PI),新算法结构简单,易于配置,具有良好的鲁棒性和网络控制性能,同时克服了大时滞给队列稳定性造成的不利影响。通过仿真表明,采用Smith-PI算法,对于限制系统振荡超调量的作用非常明显,同时能使网络具有更快的响应速度及更平稳的队列,而当网络时延增大时,算法能使网络的动态性能依然保持良好,使得缓存队列迅速收敛到稳定值。

One of the keys to assure the stability of Intemet is an efficient congestion control mechanism. Intemet congestion control system is essentially a delay system, and the RTr （Round-Trip Time） is a very important factor in designing congestion control algorithms, In terms of AQM algorithm, a new kind of congestion control algorithm （Smith-PI） is designed based on Smith predictor. With simple structure, easiness of deployment, good robustness and better performance, the proposed algorithm overcomes the negative impact on the queue stability caused by the large delay. In the end, it is illustrated by several simulations that with application of Smith-PI algorithm, the overshoot of queue can be decreased efficiently, higher responsive speed and steadier queue can be obtained, the buffer queue converge to the equilibrium point more quickly than PI algorithm and RED algorithm do in condition of HTTP disturbance or bigger overload change, and the dynamic performance can still been maintained in large-delay networks.