一种保证各流量共享瓶颈带宽的队列管理机制

随着Internet网规模的扩大，保证其应有的服务质量是网络性能研究的前沿课题。因带宽问题而造成的数据包丢失是网络服务质量的一个重要方面。总结了近10年来RED队列管理机制的各种改进方法，同时从随机排队模型出发对RED机制做了一定的性能分析，并针对它的并不能完全保证各流量共享瓶颈带宽做了称之为New_RED队列管理机制的改进,最后通过利用网络仿真工具对RED队列管理和New_RED队列管理这两种机制进行了比较：New_RED队列管理更能有效地保证各连接流量共享瓶颈带宽，而不至于在瓶颈链路上大量丢失具有拥塞控制的数据包(例如FTP流量)。

A Queue Management Mechanism with Fair Bandwidth

With the increase of Internet size, maintaining its necessary QoS(quality of service) is the forefront of network performance research. Summarized in this paper are all kinds of improved methods of RED algorithm in the past ten years. At the same time, some analyses are made of RED performance derived from random queue model and an improvement is made in maintaining fair bandwidth in all types of flows of links, resulting in New RED queue management mechanism. Finally, a comparison is made between RED mechanism and New RED mechanism by using NS. The result of simulation verifies that New RED queue management mechanism effectively maintains fair bandwidth in all types flows of links. So this mechanism doesn't drop many packets of flows which have congestion control in the bottleneck links (for instance FTP flows).