A mixed-criticality system refers to an integrated embedded system in which tasks with different criticality levels run on a shared computing platform. In the design and development of mixed-criticality systems, how to schedule tasks to ensure that high-criticality tasks are executed in time and low-criticality tasks are served as much as possible is a major problem to be studied. Existing studies tend to consider pessimistic processing strategies to ensure the schedulability of functional tasks with high-criticality requirements. However, excessive pessimistic processing can lead to waste of system resources, thereby reducing the performance of functional tasks with low-criticality requirements. In this paper, we propose an adaptive-service-level adjustment strategy for low-criticality tasks, which solves the problem of waste of resources caused by invalid compensation in the low-criticality task compensation method of flexible mixed-criticality systems. In view of the problem that the existing methods mostly use static budget allocation and static independent mode switching without considering the actual operation of the task, this paper also proposes a flexible and dynamic mixed-criticality system scheduling scheme and designs a system execution framework, scheduling algorithm, and dynamic allocation strategy of maximum execution budget, in order to reduce unnecessary redundant resource expenditures and system switching costs and to improve the performance of low-criticality tasks. Experiments show that the proposed methods are effective compared to the state-of-the-art.
Keywords: mixed-criticality system; resource allocation; service level; task scheduling.