Nonlinear time-series model for VBR video traffic

Abstract

Variable bit rate (VBR) H.261 encoded video traffic is modeled by a nonlinear time series process. A threshold autoregressive (TAR) process is of particular interest. The TAR model is comprised of a set of autoregressive (AR) processes that are switched according to the amplitude of bit rates of the time series. The TAR model is shown to capture accurately the statistical characteristics of multiplexed video sources. The buffer occupancy in a first-in first-out (FIFO) deterministic service queue using sources modeled by a TAR is compared with Markov chain based processes and it is shown that a TAR model of lower complexity yields comparable results.