Error Order of Magnitude for Modeling Autocorrelation Function of Interarrival Times of Network Traffic Using Fractional Gaussian Noise

Fractional Gaussian Noise (FGN) is a commonly used model of computer network traffic time series with Long-Range Dependence (LRD). It has been realized that FGN may not be enough to accurately model real traffic. However, quantitative evidence about this is seldom reported. To this end, this paper gives quantitative descriptions, based on processing real traffic, on the error order of magnitude for modeling autocorrelation functions of interarrival times of four types of traffic, namely, TCP, UDP, IP, OTHER, using FGN.