Partition and Scheduling of the Mixed-Criticality Tasks Based on Probability.

Lining Zeng,Cheng Xu,Renfa Li
DOI: https://doi.org/10.1109/access.2019.2926299
IF: 3.9
2019-01-01
IEEE Access
Abstract:In the mixed-criticality (MC) system, it is important to achieve better execution quality for the tasks with low criticality (LO tasks) while guaranteeing the completion of tasks with high criticality (HI tasks). To improve the execution of LO tasks, this paper studies the partition and scheduling for dual-criticality tasks on the homogeneous multiprocessor platform. First, we propose a probability MC (PMC) task model that adds a parameter of overload probability for HI tasks to the classic MC model. Second, we optimize the Earliest Deadline First with Virtual Deadline (EDF-VD) algorithm by switching the criticality of the HI tasks in sequence. Third, we analyze the possible cases in an execution period and calculate the expectation of different criticality stages based on the probability theory. Finally, we propose a partitioning algorithm based on probability (PPDC) for tasks with the PMC model. The experiments confirm the effectiveness of our calculations and show that our algorithm has better performance than the existing partition schemes.
What problem does this paper attempt to address?