Investigating Scaling Behavior of End-to-End Delay

End-to-end delay is an important QoS metric and has received much research attention. Recently, 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 analyze the scale behavior of end-to-end delay series. Based on ping series of both home and international paths, we find the 2-order correlation of delay series is analysis scale, path, and (collection) time dependent: may be biscaling or even multiscaling, persistent or anti-persistent. In addition, we observe that delay series may be multifractal at both fine and coarse scales. We also examine the source of multifractality for delay series and find the source is much complex.