基于EEMD与RBF神经网络的网络流量预测

NETWORK TRAFFIC PREDICTION BASED ON EEMD AND RBF NEURAL NETWORKS

网络流量预测对于网络性能和服务质量的提高具有重要意义。提出一种基于整体平均经验模态分解EEMD(Ensemble Empirical Mode Decomposition)与径向基函数RBF(Radial Basis Function)神经网络的预测模型,利用EEMD将长相关流量转化为短相关流量并应用RBF神经网络模型对流量数据进行建模及预测,不仅降低了算法的复杂度,而且有利于网络流量的实时预测。仿真试验结果表明,相比于自回归分数综合滑动平均模型FARIMA(Fractional AutoRegressive Integrated Moving Average Mode)、RBF神经网络模型及EMD(Empirical Mode Decomposition)与自回归滑动平均模型ARMA(AutoRegressive Moving Average Model),该模型具有更高的预测精度和良好的自适应性。

Network traffic prediction plays an important part in the improvement of network performance and service quality. The paper pro- poses a forecast model based on ensemble empirical mode decomposition （EEMD） and RBF neural network. The method translates long-term dependence traffic to short-term dependence traffic by EEMD and applies RBF neural network model to modeling and forecasting the traffic da- ta. In this way it not only reduces the complexity of the algorithm, but also does favor for the real-time forecasting of network traffic. Simulation result shows that compared with the fractional antoregressive integrated moving average model （ FARIMA）, the RBF neural network model and the EMD with autoregressive moving average model （ARMA） , the proposed model forecasts more accurately and is more adaptive.