Linear scale invariant system thorough kernels and application to self-similar wireless network traffic

In the previous work, the authors have shown that self-similar network traffic traces could be modeled by a fractional pole-zero system developed from the discrete-time continuous-dilation scaling operator effected by frequency warping. They also suggested that discrete-time linear scale invariant (DLSI) systems provide more flexible models to synthesize discrete-time self-similar processes. Furthermore, the discrete-time scaling operation and the DLSI system can be generalized through linear kernels that transform a discrete-time process to continuous time process and vice versa. This paper provides results of actual measurement of wireless LAN data and demonstrates the potential of DLSI systems in synthesizing selfsimilar traces whose statistics are consistent with that of the measured data.