近来发现 ,高速网络业务具有自相似及长相关特性 .分数噪声可描述该类业务 ,但它仅表现长相关特性 .给出了利用 FARIMA(自回归分数整合滑动平均 )模型拟合自相似网络业务的一整套方法 .该模型同时刻画了实际业务的长相关与短相关特性 .通过对实测数据的实验 ,证明了模型的优效性 Recently, it has been found that high-speed network services have self-similar and long-term correlation characteristics.Fractional noise can describe such services, but it only exhibits long-term correlation characteristics.Finally, the use of FARIMA (Autoregressive Fractional Integrated Moving Average) model to fit self-similar network services The model also depicts the long and short correlations of the real business.According to the experiment of the measured data, it proves the superiority of the model