A non-Gaussian self-similar process for broadband heavy traffic modeling

Abstract

We extend previous work on bursty network traffic characterization and management using heavy tailed self-similar processes. We revisit the proposed model for heavy network traffic, which is based on the linear fractional stable noise process to present a new algorithmic procedure for the estimation of the parameters of the model. Furthermore, we derive an asymptotic lower bound and an upper bound of the overflow probability of an ATM switch buffer.