基于周期性网络流量模型的校园网流量预测

网络流量模型是网络性能评价、网络协议设计和网络规划等的基础,然而实践证明基于泊松过程的传统流量模型并不适用于实际的网络流量.在对大量校园网络流量数据统计分析的基础上,提出一个基于周期性网络流量的网络流量模型,将网络流量分为时间相关分量和正态随机分量,并利用分布拟合检验算法加以验证,同时给出了在不同置信度下基于该流量模型的流量预测算法,从而保证对校园网络高效的管理.

CAMPUS NETWORK TRAFFIC PREDICTION BASED ON A PERIODIC TRAFFIC MODEL

Network traffic model is the basis of network performance evaluation,network protocol design and network planning.Based on the statistical analysis of a large quantity of campus networks flow,we proposed in this paper a network flow model based upon periodic network flow,the network flow can be divide into two fractions:time dependent fraction and normal random fraction.At the same time,we provided predictive algorithm under different reliability on the basis of this network flow model.