Consistent proportional delay differentiation: a fuzzy control approach

Abstract

Proportional delay differentiation (PDD) aims to maintain pre-specified packet queueing-delay ratios between different classes of traffic at each hop. Existing bandwidth-allocation approaches for PDD services rely on the assumption that average queueing delay of a class is inversely proportional to its allocated bandwidth, which is valid only in a heavy-utilized link. Consequently, in a non-heavy utilized link, they are unable to provide consistent PDD services effectively. As a remedy, in this paper, we propose a novel bandwidth-allocation approach that applies fuzzy control theory to reflect the nonlinearity between the queueing delay and the allocated bandwidth. In the approach, the bandwidth of a class is adjusted according to a set of fuzzy control rules defined over its error, the change of error, and the change of bandwidth allocation. Simulation results demonstrate that, in comparison with other bandwidth-allocation approaches, the fuzzy control approach is able to provide consistent PDD services under light, moderate, and heavy utilization conditions. They also show its robustness under various operation environments.