摘要
从多重分形树的定义出发,分析了多重分形各尺度系数的边缘分布特性及系数间自相关函数的关系,并利用得到的结论提出了一种新的MPEG-4视频流量多重分形模型(PMFM,predictable MFM)。通过对最粗尺度系数进行auto-regressive短相关预测建模,该模型具有了传统多重分形模型不具备的流量预测能力;另外,模型改进了各尺度乘子的参数获取方式,使得多重分形模型的稳定性有了显著提高。
The correlation function and marginal distribution of multipliers at each layer were investigated in multifractal multiplicative process. Based on the lemmas gained, a new multifractal MPEG-4 video traffic model (PMFM, predictable MFM)was proposed. By modeling the coarsest coefficients with the auto-regressive (AR) process, PMFM has the ability of traffic prediction that traditional MFM do not have. Furthermore, the model stability and robustness were also improved by changing the way of parameter estimation.
出处
《通信学报》
EI
CSCD
北大核心
2006年第10期44-50,56,共8页
Journal on Communications