基于排队论的实时以太网缓存队列优化算法

摘要

An optimal algorithm for a real-time Ethernet buffer queue using queuing theory is proposed.In this algorithm,frames＇ waiting delay is identified as the key factor in network delay through analyzing frames transmitting process in real-time Ethernet＇s buffer queue.The process of data frames into the buffer queue is substantially Poisson process according to stochastic process theory so that the mathematical model of transmission delay for frames queue and packet-loss based on cost is established and minimum cost in this model is considered as objective function.At last,the optimal length of queue is obtained by calculating the objective function＇s minimum value using marginal method.Experimental results show that the proposed model and optimal algorithm are correct and effective.%提出了基于排队论的实时以太网缓存队列优化算法.首先对数据帧在实时以太网缓存队列中的传输过程进行分析,确定了数据帧排队延时是影响网络延时的主要因素;然后根据随机过程理论得到数据帧进入缓存队列的过程符合Poisson分布.针对该分布模型,用排队论对数据帧排队延时及丢包建立基于通信损失代价的数学模型,以损失代价最小为目标函数,再利用边际法计算出目标函数取极值时的最佳队列长度.仿真实验验证了模型的准确性和优化算法的有效性.