基于MPEG视频流在尺度空间及小波空间的线性相关结构,提出了小波在线预测模型,并通过逆变换得到模拟业务流,从而评价网络性能。利用一阶与二阶统计方法,对预测模型在刻画流量的主要统计特性方面的性能进行仿真分析。结果表明该模型不但能较好拟合复杂业务流在各个时间尺度的概率分布特性,也能拟合其长相关特性。 Based on the linear correlation structure of MPEG video stream in scale space and wavelet space, an on-line wavelet prediction model is proposed and the simulation traffic is obtained by inverse transform to evaluate the network performance. The first-order and second-order statistical methods are used to simulate the performance of the prediction model in describing the main statistical characteristics of traffic. The results show that the model can not only fit the probability distribution characteristics of complex traffic flows at various time scales well, but also fit their long-term correlation characteristics.