Optimization of Time-Partitions for Mixed-Criticality Real-Time Distributed Embedded Systems

Abstract

In this paper we are interested in mixed-criticality embedded real-time applications mapped on distributed heterogeneous architectures. The architecture provides both spatial and temporal partitioning, thus enforcing enough separation for the critical applications. With temporal partitioning, each application is allowed to run only within predefined time slots, allocated on each processor. The sequence of time slots for all the applications on a processor are grouped within a Major Frame, which is repeated periodically. We assume that the safety-critical applications (on all criticality levels) are scheduled using static-cyclic scheduling and the non-critical applications are scheduled using fixed-priority preemptive scheduling. We consider that each application runs in a separate partition, and each partition is allocated several time slots on the processors where the application is mapped. We are interested to determine the sequence and size of the time slots within the Major Frame on each processor such that both the safety-critical and non-critical applications are schedulable. We have proposed a Simulated Annealing-based approach to solve this optimization problem. The proposed algorithm has been evaluated using several synthetic and real-life benchmarks.