摘要
计算机网络体系逐渐扩大,因此网络的服务质量和性能也急需提高。网络流量预测是网络管理的一个重要手段,研究表明网络流量具有自相似特性,在此基础上该文提出一种ARIMA预测模型。该模型首先对所生成的网络流量数据进行预处理,基于相关性与偏相关性选择ARMA模型,其次通过AIC、BIC确定阶数,利用检验后的模型进行预测,最后评估预测模型的性能。ARIMA时间序列模型能够预测非平稳数据,与传统统计模型相比,具有可忽略其他的随机变量、预测准确性更高、突发性影响较小的优点。
With the gradual expansion of computer network system,the service quality and performance of network also need to be improved.Network traffic prediction is an important means of network management.The research shows that network traffic has self similar characteristics.On this basis,this paper proposes an ARIMA prediction model.Firstly,the model preprocesses the generated network traffic data,selects ARMA model based on correlation and partial correlation,then determines the order through AIC and BIC,uses the tested model to predict,and finally evaluates the performance of the prediction model.ARIMA time series model can predict non-stationary data.Compared with the traditional statistical model,ARIMA time series model has the advantages of ignoring other random variables,higher prediction accuracy and less sudden impact.
作者
张志伟
刘立士
ZHANG Zhiwei;LIU Lishi(Shenyang Ligong University,Shenyang,Liaoning Province,110159 China)
出处
《科技资讯》
2020年第23期25-26,29,共3页
Science & Technology Information
