Multifractal Detrended Fluctuation Analysis of Network Traffic

Whether network traffic in fine scales (below about 1 second) behaves multifractality is controversial in past studies. In recent years, the multifractal detrended fluctuation analysis (MFDFA) is widely used as a robust tool to investigate the scale behavior of non-stationary time series. In this paper, we use MFDFA to review the numerical evidence of multifractality in network traffic. By carefully examining the scale behavior of four campus traffic set, we demonstrate multifractality does exist, even for UDP traffic series. In addition, we find the general two types of multifractality, (I) the multifractality due to a broad probability density function for values of the time series, and (II) the multifractality due to different long-range (time-)correlations of the small and large fluctuations, can not completely explain the multifractality in network traffic series.