This thesis is concerned with statistical time division multiplexing of variable bit rate video (moving image) sources. Its goal is to efficiently utilize a common communications channel while maintaining uniform picture quality at the receiver.
It is well known that the amount of coded data for each variable bit rate video source is correlated, and well represented by an autoregressive (AR) model.
Some characteristics of the aggregated (multiplexed) video data are derived such as batch arrival effect, arrival scheme effect, and the correlation effect. They are explained and compared with the computer simulations based on the AR model.
To analytically estimate the performance of multiplexed video coders by an approximation method, a three-state discrete-time Markov chain (DTMC) model is applied to the aggregated video sources. Analytic results for packet loss probability and packet delay time are presented for various system parameters. They are compared with the results by a continuous time Markov chain (CTMC) model of Maglaris et al., and with computer simulations.
