基于分组丢失的高带宽流鉴别算法

Algorithm to identify high-bandwidth flows based on packet drop history

提出了一种在网络发生拥塞时对高带宽流量加以鉴别的BDH-RED算法。该算法对发生分组丢弃的数据流的状态信息进行缓存,根据一个流量被丢弃的分组数正比于此流量的到达速率,在路由器计算丢弃历史记录中的分组丢弃数目,并使用组合丢弃比率,在标准TCP流量和高带宽流量共存的环境下,能有效地鉴别出高带宽流量。在仿真实验中将已有的丢弃比率与组合丢弃比率进行了比较,验证了BDH-RED算法的有效性。

A BDH-RED（ RED with Based on Drop History）algorithm was proposed to identify high-bandwidth flows in times of congestion. The algorithm will keep these flows state if some flows happen to drop packets, based on the method which packet drop number from a flow is roughly proportional to this connection arrival rate. The algorithm computes drop number in packet drop history in the routers, uses combined drop ratio, and efficiently identifies high-bandwidth flows in the coexistence environment between the normal TCP flows and high-bandwidth flows. By the comparison of existed drop ratio and combined drop ratio, the simulations valid that BDH-RED can efficiently identify high-bandwidth flows.