Joint Network Traffic Forecast with ARIMA Models and Chaotic Models Based on Wavelet Analysis

Abstract:

In the recent study of network traffic, it is shown that the traffic flow presents both periodic and self-similar characteristics. Due to these two features, the short-term forecast of network traffic cannot be accurately fit in either autoregressive integrated moving average (ARIMA) models which is suitable for linear behavior, or chaotic models which is corresponding to self-similarity characteristic. In this paper, our methodology suggests that by using wavelet multiresolution analysis, we can obtain a joint short-term network traffic prediction method and get a more precise forecast result as compared to using either ARIMA models or chaotic models. We also run simulations to show the improvement of prediction accuracy of our proposed approach.

• Periodical: Applied Mechanics and Materials (Volumes 55-57)
• Pages: 743-746
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.55-57.743
• Online since: May 2011
• Authors: Ming Ke Dong, Chen Chen, Min Hua Huang, Ye Jin
• Keywords: Chaotic Prediction, Network Traffic Prediction, Wavelet Analysis

References

[1] Konstantina Papagiannaki, Nina Taft, Zhi-Li Zhang, Christophe Diot: Long-Term Forecasting of Internet Backbone Traffic: Observations and Initial Models (IEEE INFOCOM 2003).

DOI: 10.1109/infcom.2003.1208954

[2] M. shensa, the discrete wavelet transform: wedding the A Trous and Mallat Algorithms (IEEE Transactions on Signal Processing, 1992).

DOI: 10.1109/78.157290

[3] P. Grassberger, I. Procaccia: Characterization of the strange attractors (physical Review Letters 1983).

[4] M.T. Rosenstein, J.J. Collins, C.J. Deluca: A practical method for calculating largest Lyapunov exponents from small data sets. (Physica D, 1993).

DOI: 10.1016/0167-2789(93)90009-p

[5] P. Brockwe and R. Davis: Introduction to Time Series and Forecasting, Springer, (1996).