Improved Partitioned Queue Scheduling in Multiprocessor Soft Real Time Systems

Multiprocessor scheduling in real time systems follows global queue and partitioned queue methods. Partitioned queue scheduling technique organised with dedicated queues and tasks are scheduled to corresponding processor, based on Earliest Deadline First (EDF) algorithm. EDF is an optimal technique to schedule aperiodic, periodic and sporadic tasks in Real Time Systems (RTS). Deadline of the task is a major objective of EDF and resource maintenances are negligible. It increases task migration and switching between jobs to achieve the need. Context switching reduces the system quality of services and also increases overhead. In our paper we propose D-EDF (Desirable EDF) technique with 'claim ' value µ(t), which schedules aperiodic tasks feasibly, as well as reducing the context switching and improving the efficiency of system performance apart from reducing overheads.