Abstract: The TCP source congestion control disciplines are introduced to the router congestion control problem, and an Active Queue Management(AQM) algorithm called Expecting Rate Fuzzy Update(ERFU) is proposed. It considers both the afferent flow rate and the inside load, and directly updates the dropping (or marking) probability of each afferent packet based on the TCP source congestion control disciplines to adjust the queue length quickly converging to designated aim. It needs low requirement and performs superiorly and proves to have better performance than ERED.

Key words: congestion control, Active Queue Management(AQM), source control disciplines

摘要： 将TCP源端的拥塞控制规律引入路由器端的拥塞控制问题，提出一种高性能的主动队列管理算法ERFU，综合考虑路由器输入流的流量速率与内部负载程度，无需设置多样的参数，直接依据TCP源端的拥塞控制规律，更新数据包的丢弃(标记)概率，具有较低的算法需求和优越的算法性能。通过仿真与ERED相比较，证明ERFU在随机流、突发流2种网络状况下均具有更优良的性能。

关键词: 拥塞控制, 主动队列管理, 源端控制规律

CLC Number: 

• TP301.6