A Semi-Partition Algorithm for Mixed-Criticality Tasks in Multiprocessor Platform

Abstract

In the area of mixed criticality scheduling, a hot topic is the partition and scheduling in multiprocessor platform. In this paper, we propose a semi-partition mechanism to optimize the scheduling of mixed-criticality tasks in homogeneous multiprocessor platform. The mechanism firstly partition the HI tasks and LO tasks separately based on utilization which can guarantee the reliability of system. Then in runtime, to make better use of computing resource, the mechanism allows conditional migration of LO tasks. The experiments show that our mechanism has better performance in scheduling the mixed-criticality tasks.