基于映射矩阵的网络流量分析预测模型

Network traffic analysis and forecasting model based on mapping matrix

通过对用户行为分析,发现IP数据流具有平稳性、自相关性等特点,提出基于映射矩阵流量预测模型,并与线性模型AR、ARIMA和非线性基于反馈神经网络BP模型、Elman神经网络作对比,试验结果证明,映射矩阵模型,比现有模型具有预测精度高、收敛快等特点。

By analyzing user behavior, this paper finds that the IP data stream has smooth and self-correlation characteristics, pro- poses a prediction model based on the mapping matrix flow in comparison with the linear AR, RIMA models and nonlinear BP model based on feedback neural network, Elman neural network. The test results demonstrate that the mapping matrix model has high prediction accuracy and fast convergence characteristics compared with existing models.