A robust adaptive effective bandwidth allocation scheme

Efficient bandwidth allocation has been a popular research topic during the last few years. It is a challenging issue to provide guaranteed quality of service (QoS) for network applications while still obtaining high network utilization. As a promising approach to achieve tradeoff between network utilization and the provisioning of QoS, the concept of effective bandwidth has been widely accepted. However, it requires a full characterization of the underlying process to calculate its effective bandwidth, which is not trivial. It is also well known that the pure effective bandwidth allocation is conservative. To bypass modeling the underlying traffic and overcome the conservative nature of effective bandwidth, we propose a robust adaptive effective bandwidth allocation (AEBA) algorithm. We study the performance of the robust AEBA method under the dynamic weighted round-robin (DWRR) scheduling with a set of simulations using both self-similar traffic and traditional Poisson traffic as input. The simulation results show that our approach allows different QoS requirements to be satisfied while effectively exploiting the statistical multiplexing gain occurring among multiple traffic classes at the same time. The simulation results also show that our approach is robust in that it does not need any assumptions about the underlying traffic.