Abstract: Active queue management plays an important role in congestion control of network. Whereas most algorithms of the active queue management such as Random Early Detection(RED) have the problem of strongly relying on parameters. Aiming at the problem that the average queue length is unable to fully reflect the network congestion in the algorithm of RED, this paper puts forward an improved RED algorithm with the average queue and network load. This algorithm can adjust the probability of the pack loss adaptively according to the variety of network load so that it will be more congruous to the actual status. The validity of the algorithm is finally proved by performance analysis via emulate.

Key words: congestion control, active queue management, Random Early Detection(RED)

摘要： 主动队列管理是实现网络拥塞控制的重要技术，但是多数主动队列管理算法如随机早期检(RED)都存在对参数依赖性强的问题。针对RED算法中平均队列长度不能完全反映网络拥塞状况的问题，该文结合平均队列长度和网络的负载，提出一种改进的RED算法。该算法能根据网络负载的变化，自适应地调整丢包的概率，使它更符合网络的实际状况。通过仿真进行了性能分析，证明了算法的有效性。

关键词: 拥塞控制, 主动队列管理, 随机早期检测

CLC Number: 

• TP393