Abstract: This paper investigates RED algorithms’ elements and limitation and introduces a new nonlinear high order RED algorithm which employs a high-order function to mark or discard fragments with lower and higher probability near low and high threshold respectively. The enhanced RED algorithm is also implemented in the network simulator NS2, the results of series simulation experiments show that the algorithm improves performance, stableness and reliability of network.

Key words: Congestion control, Random early detection (RED), Active queue management

摘要： 分析了RED及其改进算法的原理和局限性，提出了一种非线性高阶RED拥塞控制机制。该算法利用一个高阶分组丢弃函数，在下限阈值附近以较低的概率标记丢弃分组，在上限阈值附近迅速提高分组的标记丢弃概率。并对网络仿真器NS2进行扩展，通过系列仿真实验验证了改进算法有效提高了网络性能。

关键词: 拥塞控制, 早期随机丢弃, 主动队列管理

CLC Number: 

• TP393