RED-PD算法的仿真研究

Simulation Study of RED with Preferential Dropping

随机早期检测(RED:Random Early Detection)是IETF推荐的一种基于路由器有效的主动队列管理算法,但是在某些情况下,一些数据量很大的数据流会大量占用带宽,从而导致了各流量之间带宽分配的不公平性,甚至产生拥塞崩溃。对M ahajan提出的一种基于RED分组丢弃历史的AQM(Active Queue M anagem ent)算法RED-PD(RED w ith Preferential D ropp ing)进行了深入研究。该算法通过对被检测出的高带宽流的数据包采用提前丢弃的策略,实现带宽分配的公平性。通过仿真发现了RED-PD算法的自适应性,表明此算法有更高的理论和实用价值。

RED （Random Early Detection） is an efficient AQM （Active Queue Management） algorithm based on routers by IETF recommendation. But in some network environment, RED provides little protection from high- bandwidth flows that take much wide bandwidth, which can result in extreme unfairness among per-flow, even congestion collapse. Mahajan proposed an AQM algorithm, RED-PD （RED with Preferential Dropping）. Based on history of packet drops from RED queues, the algorithm provides the fairness among per-flow through ahead dropping packets from high bandwidth flows which have been tested. Simulation results show that RED-PD is self-adaptable, and this algorithm is useful theoretically and practically.