一种基于改进的模糊Smith控制主动队列管理算法  

An Improved Fuzzy-Smith Approach for Active Queue Management

作　　者：何凌[1] 张楠楠[1] 杨木易[1] 井元伟[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《东北大学学报（自然科学版）》2007年第12期1673-1676,共4页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(60274099);国家高技术研究发展计划项目(2004AA412030)

摘　　要：针对网络中传输延时给拥塞控制带来不利影响,而现有算法过于依赖精确模型的弱点,提出将模糊控制与改进的Smith预估补偿相结合的设计用于与TCP连接的主动队列管理算法.通过改进的Smith预估补偿器对时延进行预估补偿,使得对TCP的拥塞控制更加及时;而模糊控制器无需被控对象的精确数学模型即能实现良好的控制,可以克服Smith预估补偿依赖精确模型的缺点.两者的结合对于时滞网络的控制可以达到优势互补的作用.最后通过仿真验证了方法的有效性.The transmission delay brings about disadvantages of congestion control in network, while the relevant existing algorithm depends on the exactitude of model excessively. A new active queue algorithm is thus proposed by improving Smith estimate compensator in combination with fuzzy control, which is link up with TCP. The improved Smith estimate compensator can control more timely the TCP congestion, and fuzzy controller can get rid of an exact model for the object to control. In this way, the advantages of Smith estimate compensator and fuzzy controller may complement each other. The results of numerical example show the validity of such a tactical combination.

关 键 词：拥塞控制 模糊控制 Smith预估算法 大滞后 主动队列管理 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]