Earliest Due First Scheduling for Application-Level QoS Delivery

The Quality of Service (QoS) of real time applications is aaected in a rather direct manner by the timely delivery of application-level data units which are typically mapped into several ATM cells. As a consequence, it would be reasonable to associate a common deadline with all the cells in one data unit (batch) and develop a scheduling policy that aims to maximise the number of data unit that are delivered by their respective deadlines. A batch-based Earliest Due First scheduling policy is proposed and analysed in this paper under periodic batch arrival processes which are typical of continuous media streams. Depending on whether partial batch service is performed or not, two variations of this scheduler are considered. It is shown that the proposed policies can be easily implemented in terms of a First Batch In-First Batch Out (FBI-FBO) and a Partial FBI-FBO schedulers respectively. Numerical results are presented and the induced batch and cell loss rates are compared against those induced by the standard FIFO scheduler as well as a FIFO scheduler which drops expired cells.