1st International ICST Conference on Communications and Networking in China

Research Article

Multi-scale Internet Traffic Prediction Using Wavelet Neural Network Combined Model

  • @INPROCEEDINGS{10.1109/CHINACOM.2006.344786,
        author={Chen  Di and Feng  Hai-liang and Lin Qing-jia and Chen  Chun-xiao},
        title={Multi-scale Internet Traffic Prediction Using Wavelet Neural Network Combined Model},
        proceedings={1st International ICST Conference on Communications and Networking in China},
        publisher={IEEE},
        proceedings_a={CHINACOM},
        year={2007},
        month={4},
        keywords={wavelet transform artificial neural network prediction timescale.},
        doi={10.1109/CHINACOM.2006.344786}
    }
    
  • Chen Di
    Feng Hai-liang
    Lin Qing-jia
    Chen Chun-xiao
    Year: 2007
    Multi-scale Internet Traffic Prediction Using Wavelet Neural Network Combined Model
    CHINACOM
    IEEE
    DOI: 10.1109/CHINACOM.2006.344786
Chen Di1, Feng Hai-liang1, Lin Qing-jia1, Chen Chun-xiao1
  • 1: School of Information Science and Engineering, Shandong Univ, Jinan 250100, Shandong, China

Abstract

Internet traffic belongs to non-stationary time series, wavelet transform can decompose non-stationary time series into several stationary components. In this paper, we decompose the Internet traffic with wavelet first, and then apply two different artificial neural network (ANN) architectures, linear neural network (LNN) and Elman neural network (ENN), to predict the components. The LNN predicts the linear data, whereas the ENN predicts the nonlinear data. To enhance the prediction accuracy and merge the traffic characteristics captured by individual models, the outputs of the individual ANN predictors are combined using three networks respectively. They are back propagation neural network (BPNN), LNN and ENN. The problem of one-step-ahead traffic prediction at different timescales is considered. The results indicate that the proposed combined model outperforms the individual models and the wavelet transform improves the performance further. The results also show that the prediction performance depends on the traffic nature and the considered timescale