【摘要】 为了解决DiffServ网络中带宽利用的不公平性 ,该文提出了一种自适应的数据包标记算法AFM(AdaptiveFairMarker) .与现有的标记算法相比 ,它有两个显著的不同 :( 1)增加了一种带宽估计机制 ,对网络中可使用的带宽进行动态估计 ,并将所估计的带宽以按比例的方式公平地分配给各个汇聚流 ;( 2 )对TCP协议作了一个微小而又非常有效的改进 ,在控制TCP拥塞窗口的同时尽量避免TCP协议的AIMD机制 .该文通过仿真试验对算法进行了验证 ,结果证实AFM算法比其它几种算法具有更好的公平性更多还原

【Abstract】 Recently, the study of DiffServ is increasingly enriched. Due to its simplicity and extensibility, DiffServ has been widely considered as a promising model. Researchers found unfairness in DiffServ networks. To solve the problem of unfairness, this paper proposes an adaptive fair marker, which is called AFM. Compared with other schemes, AFM has two prominent difference: (1) It has a built in bandwidth estimation mechanism, with which, AFM can estimate the available bandwidth dynamically, and reallocate the bandwidth among the aggregates proportionally to their target rate; (2) A tiny and effective modification has been made on the TCP protocol, with which, we can control the congestion window and avoid or decrease the times of the AIMD mechanism. Authors validate this marking algorithm through simulation experimentation, and compare the simulation results with those of some other marking algorithms, results show that the AFM algorithm has better fairness than other algorithms.更多还原