中文摘要：

根据网络流量的非线性特征，及分块分形插值函数能模拟分形信号的优点，对震荡类型不同的两类流量重构，发现其对不同尺度流量拟合误差都很小，提出将该方法用于在很大尺度范围内的网络流量多尺度结构研究。进一步研究采用分块FIF重构的信号与原信号在多分形、能谱等分形特性和统计特征方面的保持能力；通过对信号FIF重构过程的分析，讨论了分块FIF构造过程与重构信号的多分形多尺度结构间的关系。提出分块FIF方法是基于流体技术研究网络流量多分形多尺度结构的有效工具。

英文摘要：

Aiming at the nonlinearity of network traffic, piecewise FIF is used to reconstruct two kinds of traffics different on oscillating frequencies by using its advantages on fitting fractal signals. Experiments show that piecewise FIF could fit irregular network traffic well. So it is proposed to simulate signals in a larger, scope and analyze the multiscale structure of network traffic. The multi-fractal, power spectrum and statistical properties of fitting data are compared with orginal signals. The relationship between FIF and properties of fitting data is discussed according to the process of signal reconstructing. It is showed that the piecewise FIF is an effective tool to study the multi-fractal multiscale structure of network traffic based on fluent technologies.