Abstract: In this paper, first, we used Multi-fractal Wavelet Model (MWM) to characterize the MPEG-4 video traces. The model used the Haar wavelet transform and put a constrain on the wavelet coefficients to guarantee positivity. Then, we used first-order, second-order statistical methods and multi-fractal analysis ways to study the main characteristics of the simulated traces . These characteristics include pdf function, LRD, burstiness and the multi-fractal feature of the video traces. At last, we evaluated the model’s performance. Our findings are that: MWM is a simple and useful model for characterizing MPEG-4 video traces. I t can characterize the global statistical characteristics and the multi-fractal feature of the MPEG-4 video traces perfectly. But this model has its default; it exhibits a poor performance when it is applied to the traces with strong burstiness.

摘要： 本文采用多分形小波模型（MWM）对MPEG—4视频流进行仿真，其母小波采用Haar小波，并利用一阶统计、二阶统计和多分形分析的方法，对模型在刻画流量的主要统计特性（如概率密度分布、LRD特性、突发性和多分形特征）方面的性能进行了评估，其结果为：MWM模型是一个简单的和有实用价值的MPEG—4视频流的仿真模型，它可以很好地刻画出流量的全局统计特性和多分形特征，但此模型也有缺陷，对于具有高突发性的视频流量的刻画能力较弱，需要我们在以后的研究工作中加以完善。