摘要
随着科学技术的快速发展,互联网已经进入大数据时代,网络业务的类别和数据量急剧增加,导致网络流量数据的分析方法也随之发生了明显改变。网络流量预测是网络流量分析的一种重要方法,其分析结果不仅可以指出网络流量的未来趋势,而且通过对流量预测值的分析能够提前发现网络异常。传统的网络流量模型在如今的流量分析与预测中存在一定限制,因此探索新的网络流量建模及预测方法势在必行。通过建立自回归积分滑动平均(ARIMA)模型对贝尔实验室提供的BC-Oct89Ext实测流量数据进行网络流量预测,该模型对于部分非平稳网络流量数据的分析和预测的准确性较高,网络流量预测结果可以对网络趋势进行相应的网络资源调整,保证网络业务的服务质量与服务效率,对网络业务的快速发展具有重要意义。
With the rapid development of science and technology, the Internet enters the era of big data,and the types and data volume of network services increases sharply, resulting in a significant change in the analysis methods of network traffic data. Network traffic prediction is an important method for network traffic analysis, and the analysis results can indicate the future trend of network traffic and detect network anomalies in advance by analyzing traffic prediction values. Traditional network traffic models have certain limitations in today's traffic analysis and prediction, therefore, it is imperative to explore new network traffic modeling and prediction methods. An ARIMA(autoregressive integral moving average) model is established to predict network traffic for the BC-Oct89 Ext measured traffic data provided by Bell Labs. This model has high accuracy for the analysis and prediction of some non-stationary network traffic data, and the network traffic prediction result may serve as a basis for adjusting the network resources of the network trend, ensuring the service quality and service efficiency of the network service, and all this is of great significance to the rapid development of network services.
引文
[1]Mark C,Balachander K.Internet Measurement Infrastructure,Traffic and Applications[M].Chi chester:John Wiley&Sons Ltd,2006.
[2]杨家海,吴建平,安常青.互联网络测量理论与应用[M].北京:人民邮电出版社,2009.YANG Jia-hai,WU Jian-ping,AN Chang-qing.The Theory and Applications of Internet Measurement Beijing:The People’s Posts&Telecom Press,2009.
[3]Box G,Jenkins G.Time Series Analysis,Forecasting,and Control[M].San Francisco:Holden Day Press,1970.
[4]FAN Jing-qing,Yao Qi-wei.Nonlinear Time Series Nonparametric and Parametric Methods[M].Beijing Science Press,2006.
[5]Akaike,H.Cononicial Correlation Analysis of Time Series and the Use of an Information Criterion,in Systems Identification:Advances and Case Studies[M].New York:Academic Press,1976.