DARTS: DynAmic Real-time Task Scheduling

Abstract

Real-time systems often involve time critical control tasks in which their correctness depends not only on the functionality but also on timeliness. In order to guarantee the timely-correctness, real-time scheduling has been studied extensively. The main problem of multiprogramming scheduling on a single processor is that an optimum fixed priority scheduler has a least upper bound to processor utilization, which is around 70 percent for large tasksets, in contrast, full processor utilization can be achieved by dynamically assigning priorities. In this paper DynAmic Real-time Task Scheduling (DARTS) algorithm is proposed, which is based on dynamic utilization and assigns higher priority to a task with the highest utilization with regard to its laxity. Eventually, this method is extended for multiprocessor systems and we demonstrate that not only does DARTS algorithm have better utilization than existing global EDF schedulability tests, but it also has significantly well outputs in total utilization less than 90 percent.