一种有效的网络流量预测算法

Efficient Algorithm for Network Traffic Prediction

为了对网络流量做定量研究,提出一种基于稳态队列长度的新型网络实际流量预测算法(Prediction algorithm based on Queue length of Steady state for FARIMA model,PQSF).该算法首先利用基于稳态队列长度的乘积解理论推导节点数据包的排队情况,计算出存在失效节点时流量平均对长的数学公式,并结合FARIMA模型建立预测方法,最后通过网络仿真对PQSF算法进行验证.实验结果表明,该算法具有较好的适应性.

In order to make quantitative research on network traffic,this paper puts forward a kind of prediction algorithm based on the queue length of steady state for FARIMA model,PQSF. The algorithm firstly uses the theory of product form solution based on the stable length of queue to derive node packet queue,then calculates the average flow rate for long mathematical formula when there are failure nodes,and establishes the prediction method combined with FARIMA. Finally,the paper validates the PQSF algorithm through the network simulation. The results show that the algorithm has comparatively good adaptability.