一种用于拥塞网络节点缓存队列长度控制方法

One for Congestion Nodes Cache Queue Length Control Method

控制拥塞网络节点的缓存队列长度,在提高网络利用率方面具有重要意义。由于拥塞网络中传送的节点数据太多,使得节点缓存队列长度受到限制。采用传统的控制方法,当存储节点受限时对节点缓存队列进行剪裁来增加传输速率,难以有效的对节点缓存队列进行控制。现提出一种量子粒子群算法,并应用到拥塞网络节点缓存队列长度的控制中。利用概率接纳算法对网络的拥塞状态进行检测,以检测结果为依据对消息进行接受与丢弃处理,采用PI控制器建立PI主动队列管理模型,获取当前时刻拥塞网络节点缓存队列数据包丢弃率,引入量子粒子群算法通过对模型比例系数和积分系数的优化,实现对拥塞网络节点缓存队列长度控制。仿真结果表明,采用改进的控制方法可以有效的降低节点丢包率,提高网络吞吐率,降低控制误差。

A quantum particle swarm optimization algorithm is proposed and used to control buffer queue length of nodes in congestion network. Firstly,the probability admission algorithm is used to detect network congestion status.The message is accepted or abandoned according to the detection result. Then a PI management model of active queue is built by using PI controller to obtain data package loss rate of node buffer queue at present. Finally the quantum particle swarm optimization algorithm is introduced to achieve the length control through optimization of proportion coefficient and integral coefficient. The simulation results show that the modified method can reduce packet loss rate of node. It can improve network throughput rate and make the controlling error decline.