Energy-Efficient Scheduling of Real-time Tasks with Abortable Critical Sections

In this paper, we propose a ceiling-based scheduling algorithm, called conditional abortable task scheduling algorithm (CATSA), to schedule real-time tasks with abortable critical sections in a non-ideal dynamic voltage scaling processor. To minimize the energy consumption, the CATSA uses the lowest possible processor speed for task executions such that all the timing constraints of tasks are satisfied. The aborting of a critical section will increase the system utilization because it has to be re-executed from the beginning of the aborted critical section. Hence a critical section will be aborted only if the re-execution of the aborted critical section is more energy-efficient than the blocking of higher-priority tasks.