№2, 2020


Ramiz H. Shikhaliyev

Accurate prediction of the traffic volume of computer networks (CN) for both the long and short term plays a crucial role in monitoring, as well as in the effective management of the optimal use of available network resources. Typically, more experienced network administrators intuitively predict the traffic volume of the CN, however this is completely unacceptable for the administration of modern, large and complex CNs. Therefore, more accurate methods for predicting the traffic volume of the CN should be developed using machine learning methods that will help the CN administrators effectively plan and optimally manage the use of available network resources. This article proposes a method for short-term CN traffic volume prediction based on the CART (Classification and Regression Trees) model. The essence of the method is that, using decision trees, the previous sets of traffic volume states are classified according to the set of traffic volume state patterns and a linear regression model is constructed corresponding to each class. The method allows predicting the future CN traffic volume state by clustering vectors of current traffic volume states according to the most suitable previous patterns and using regression. Thus, the task of CN traffic volume short-term prediction is to determine the vectors of the current traffic volume states and a regression model for forecasting. To assess the accuracy of the method, the MASE (Mean Absolute Scaled Error) is used. The proposed method allows predicting the traffic volume of the CN for short periods of time, for example, for weeks, days, hours and seconds. The results of short-term forecasts can be used to improve QoS (Quality of Service), prevent overload of communication channels, optimal control of the available CN resources, etc (pp.124-133).

Keywords: computer networks, network traffic, traffic prediction, CART model, classification, decision tree.
  • Zhani M.F., Elbiaze H. Analysis and Prediction of Real Network Traffic // Journal of networks, 2009, vol. 4, no. 9, pp. 855−865.
  • Breiman L., Friedman J.H., Olshen R.A., Stone, C.J. Classification and Regression Trees. Wadsworth International Group, Belmont CA, 1984, 368 p.
  • Hoong N.K., Hoong P.K., Tan I.K.T., Muthuvelu N.M., Seng L.C. Impact of Utilizing Forecasted Network Traffic for Data Transfers / IEEE 13th International Conference on Advanced Communication Technology, 2011, pp.1199−
  • Hoong P.K., Tan K.T., Keong C.Y., BitTorrent Network Traffic Forecasting With ARIMA // International Journal of Computer Networks & Communications, 2012, vol.4, no.4, pp. 143−156
  • Sadek N., Khotanzad A. Multi-scale High Speed Network Traffic Prediction Using K-Factor Gengendaue ARMA Model / IEEE International Conference on Communications, 2004, pp. 2148−2152.
  • Yu Y., Wang J., Song M., Song J. Network Traffic prediction and result analysis based seasonal and ARIMA and Correlation Coefficient / IEEE International Conference on Intelligent System Design and Engineering Application, 2010, vol.1, pp. 980−983.
  • El Hag H.M.A., Sharif S.M. An Adjusted ARIMA Model for Internet Traffic / IEEE AFRICON Conference, 2007, 6 p.
  • Anand N.C., Scoglio C.S., Natarajan B. GARCH Non-Linear Time Series Model for Traffic Modeling and Prediction / IEEE Network Operations and Management Symposium, 2008, pp. 694−697.
  • Park C., Woo D-M. Prediction of Network Traffic by Using Dynamic Bilinear Recurrent Neural Network / IEEE Fifth International Conference on Natural Computation, ICNC 2009, Tianjian, China, 14-16 August 2009, pp. 419−423.
  • Chabaa S., Zeroual A., Antari J. Identification and Prediction of Internet Traffic Using Artificial Neural Networks // Journal of Intelligent Learning Systems and Applications, 2010, vol. 2, no. 3, pp. 147−155.
  • Junsong W., Jiukun W., Maohua Z., Junjie W. Prediction of Internet Traffic Based on Elman Neural Network / IEEE Chinese Control and Decision Conference, 2009, pp. 1248−1252.
  • Chabaa S., Zeroual A., Antari J. ANFIS Method for Forecasting Internet Traffic Time Series / Mediterrannean Microwave Symposium, 2009, pp. 1−4.
  • Zhou B., He D., Sun Z. Traffic predictable based on ARIMA/GARCH Model / IEEE 2006 2nd Conference on Next Generation Internet Design and Engineering, 2006, pp. 200−207.
  • Zeng D., Xu1J., Gu J., Liu L., Xu G. Short Term Traffic Flow Prediction Using Hybrid ARIMA and ANN model / IEEE Workshop on Power Electronics and Intelligent Transportation System, 2008, pp. 621−625.
  • Rutka G., Network Traffic Prediction using ARIMA and Neural Networks Models // Elektrotechnika, 2008, vol. 84, no. 4, pp. 53−58.
  • Iqbal M. F., John L. K. Power and Performance Analysis of Network Traffic Prediction Techniques / IEEE International Symposium on Performance Analysis of Systems & Software, 2012, pp. 112−113.
  • Guang C., Jian G., Wei D. A Time series Decomposed Model of Network Traffic // Springer, 2005, pp. 338−345.
  • Hyndman R., Athanasopoulos G., Forecasting: principles and OTexts, 2014. https://books.google.com
  • Theyazn H. H. Aldhyan, Manish R. Joshi Intelligent Time Series Model to Predict Bandwidth Utilization // International Journal of Computer Science and Applications, 2017, vol. 14, no. 2, pp. 130−141
  • Jung S, Kim C., Chung Y. A Prediction Method of Network Traffic Using Time Series Models, ICCSA 2006, pp. 234−243.
  • Hyndman R. J., Koehler A. B. Another look at measures of forecast accuracy // International Journal of Forecasting, 2006, vol. 22, no. 4, pp. 679−