【摘要】 提出了一种基于ARMA(自回归滑动平均)模型的经验模式分解预测自相似网络流量的方法,进行了理论证明和仿真验证。结果表明,经验模式分解对长相关流量有去相关的作用,采用ARMA模型即可对自相似网络流量准确刻画,不但降低了算法的复杂度,而且预测精度高于径向基函数神经网络的预测精度。更多还原

【Abstract】 A novel method based on empirical mode decomposition(EMD) and ARMA was proposed to model and fore-cast self-similar networking traffic.The results demonstrate that EMD had the function of getting rid of the long range dependence(LRD) in traffic data.Therefore,the self-similar traffic processed by EMD could be modeled and predicted well by using ARMA which was a short range dependent(SRD) model.Moreover,the complexity of the proposed method was reduced sharply and the prediction precision was higher than radial basis function neural network.更多还原