The Hurst Index Estimation and Network Performance Analysis Based on ON/OFF Model

ON/OFF model is the most attractive self-similar traffic generating methods due to a clear physical meaning of the self-similarity/long-range dependence. The Hurst is the most important parameters for the simulated traffics to be estimated. However, the estimated Hurst values often change with the sampling frequency. In this paper, firstly, we under-sample and over-sample the network traffic simulated based on the ON/OFF model, and show the phenomenon that how the estimation value change with the sampling frequency. Finally, we analyze the correlations between packet loss rates, network relay, relay jitter and buffer size.