Parallel Real-Time Scheduling

Jing Li, Kunal Agrawal, Chenyang Lu

Abstract: This chapter of the survey provides a basic introduction to real-time scheduling for parallel tasks on multicore platforms and gives an overview of the relevant results. Parallel real-time tasks are the tasks that can utilize multiple cores at the same time to complete more computation with the same real-time constraints than sequential tasks. For parallel real-time systems, researchers have studied three types of scheduling strategies: decomposition-based scheduling, global scheduling, and federated scheduling. The survey summarizes the results of these three types of scheduling strategies, including results for different parallel task models, different response time analyses, and theoretical bounds, as well as system implementation and empirical evaluations. In addition, the survey also covers some work for mixed-criticality systems that have parallel tasks.