MPEG视频流小波在线预测模型  

Modeling MPEG Video Traces Using Wavelet On-line Prediction Model

作　　者：霍炎[1] 荆涛[1] 冯玉珉[1] 张德珍[1] 

机构地区：[1]北京交通大学电子信息工程学院多媒体实验室,北京100044

出　　处：《系统仿真学报》2008年第13期3351-3354,3362,共5页Journal of System Simulation

基　　金：国家自然科学基金资助项目(60472034)

摘　　要：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。The Wavelet On-line Prediction Model based on the linear correlation structure of scaling space and wavelet space in MPEG video trace was proposed. The simulated traces were generated by inverse wavelet transform for evaluating network performance. First-order and second-order statistical methods were used to study the main characteristics of the simulated traces. It is verified by the simulations that this model can capture the probability distribution and the long-range dependence of the MPEG video trace very well.

关 键 词：MPEG视频流 线性相关结构 预测模型 小波 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]