一种分阶段自适应RED／ECN参数模型

TCP拥塞控制技术是IP网络性能的研究重点之一，目前已经出现了多种改进方案，然而采用这些方案后，TCP连接仍然面临大量的包丢失，尤其是在网络拥塞时。为此IETF提出了IP网络的拥塞控制问题并建议在网络中采用主动队列管理机制。RED／ECN算法是一种主动队列管理算法，它具有很多优点，但算法对其静态参数的依赖性很大，没有充分考虑链路中复用的活动连接数，因而不能很好地适应网络负荷的变化。为此，提出了一种分阶段动态调整RED／ECN算法参数的模型（GARED），通过仿真实验说明该模型改进了RED／ECN算法的性能，能够有效地降低丢包率，保证网络链路的高利用率。

A GRADING ADAPTIVE MODEL OF RED/ECN PARAMETERS

TCP congestion control mechanisms are one of the focuses of numerous studies of the IP network performance and have undergone a number of enhancements. But TCP connections still experience alarmingly high loss rates, especially during the time of congestion. The IETF has proposed the solution by deploying active queue management mechanisms in the routers, such as RED/ECN algorithm. But it is found that its effectiveness is highly dependent on its static parameters and has no sufficient consideration about the number of connections multiplexed over the link, so it can not adapt to the changes of network load effectively. To alleviate this problem, a grading adaptive model of parameters is proposed. The simulation results indicate the model can effectively reduce packet loss while maintaining high link utilization under various network conditions.