Dynamic management of real time QoS according to (m, k)-firm

Abstract

A model of real time QoS (quality of service) according to constraint of (m, k)-firm offers more flexibility and a better use of resources compared to guaranteed QoS. It opens new prospects for the implementation of the adaptive management mechanisms of QoS in real time networks and systems. To take into account this new type of constraint during the scheduling of access to the resources, the scheduling algorithms of QoS developed for the hard real time system (for example EDF, FP) and for the hard real time network (for example WFQ) must be extended. This article first presents a state of the art on the principal scheduling algorithms developed for the real time (m, k)-firm guarantee, then two algorithms of our recent work: E-DBP and (m, k)-WF/sup 2/Q. We also define the notion of the gracious degradation of QoS according to (m, k)-firm and discuss its application to dynamic management of QoS in the networks.