Abstract: A Weight Sum-Based Random Early Detection (WHS-RED) algorithm was proposed to deal with network congestion. WHS-RED algorithm used the rate of change of queue lengths in networks to drop network packets. The mathematical model was established. Simulation results prove that the proposed algorithm reduces packet loss rate and bandwidth-delay, while maintaining network stability and reducing the jitter of the queue, which is better than that of the original RED algorithm.

Key words: queue, Random Early Detection (RED), congestion control, Exponentially Weighted Moving Average （EWMA）, average queue length, marking probability

摘要： 为控制网络拥塞，提出一种改进的随机早期检测（RED)算法——基于加权和的随机早期检测（WHS-RED）算法。算法的基本思想是利用网络中队列长度及其变化率的加权和，动态地控制网络丢包率。建立了数学模型，并给出了计算机仿真分析结果。实验证明改进算法减小了网络丢包率和带宽延迟，在维持网络稳定性和减小队列的波动性方面优于原RED算法。

关键词: 队列, 随机早期检测, 拥塞控制, 指数加权滑动平均, 平均队列长度, 标记概率