A method of multifractal traffic generation with appropriate short-term distribution using nonlinear filtering

This paper proposes a method of generating network traffic patterns needed in stress testing of Web sites. This method mainly consists of two stages: in the first stage, from a positive root signal, a signal pattern with multifractality and a long-tail distribution is synthesized using wavelets. In the second stage, the short-term distribution characteristics of the pattern generated at the first stage are adjusted by decomposition into low-frequency and high-frequency components by nonlinear filtering and changing the ratios of their two components in order to obtain the appropriate short-term distribution, so as to finally obtain network traffic that also has an appropriate long-range dependence. Simulation results are also presented for validation of this proposed method. © 2005 Wiley Periodicals, Inc.