Effective bandwidths under dynamic weighted round robin scheduling

We develop a framework of using effective bandwidths under dynamic weighted round robin scheduling to study the statistical quality of service assurance issue in self-sizing networks supporting Differentiated Service. A traffic measurement-based adaptive effective bandwidth allocation algorithm aiming at improving the performance of effective bandwidths is proposed. We evaluate our proposed mechanism with a set of simulations that use Poisson and Markov modulated Poisson process sources as input. The simulation results show that the adaptive effective bandwidth allocation allows different quality of service requirements to be satisfied at the same time while overcoming the conservative nature of the pure effective bandwidth allocation.