基于小波变换和时间序列的网络流量预测模型

Network traffic prediction based on wavelet transform and time series model

随着互联网规模和应用的扩大,网络数据流量呈现出复杂多分形性的特点,针对这个特性,构建了基于小波分析和ARMA模型的网络流量预测模型,用Mallat算法将原始流量数据分解为4个分层数据,对各层数据用ARMA模型进行预测,再将各层预测数据重组为预测的网络流量.采用真实数据进行仿真的试验表明,基于小波分析和ARMA相结合的网络流量预测模型的预测结果具有较高的准确度,并在网络管理和优化中具有重要实用价值.

With the increasing number of network people and internet application, network traffic data become the characteristic of muhi-fractal. Focusing on the character, a hybrid model based on the wavelet transform and ARMA model is proposed. The original data are transferred into 4 layers data by Matlab algorithm, and ARMA models apply in each layers data to predict the future data. Then we reconstruct them into the predicted network data. By applying the data collected in a real network environment, the simulation experiment on the hybrid model is taken. The experiment result shows that the hybrid model has higher accuracy on network traf- fic predication and is practical on network management and optimization.