EN FR
EN FR


Section: New Results

Multicore processor graph tasks scheduling

Due to widespread of multicore processors on embedded and real-time systems, we concentrate our work on the study of the schedulability of real-time tasks with precedence constraints on such processors. We consider preemptive fixed-priority scheduling policies. First, we have proposed a response time analysis for directed acyclic graphs task model with non-probabilistic execution time and preemptive fixed-priority scheduling policy [10]. Our response time analysis improves importantly the state of the art analyses, while allowing scalable extensions for response time analysis of tasks with worst case execution times described by probability distributions. We extend this response time analysis to similar task model with probabilistic worst case execution time with the advantage of providing efficient results also for task model with non-probabilistic worst case execution times. Our response time analysis is based on iterative equations which offer run-time enhancement compared to existing work [21] requesting the resolution of complex MILP optimization problem. In addition, we have defined priority on sub-task level enhancing the schedulability and reducing the worst-case response time. The proposed priority assignment algorithm is adapted for the studied task model and it outperforms several state-of-the art methods. We have also proposed a partitioning heuristic that assigns each sub-task to a given core. This heuristic takes into consideration communication delays between sub-tasks inside the same graph in order to minimize the communication while balancing different cores load and maximizing possible parallelism. The proposed heuristics and response time analysis (RTA) are validated on randomly generated task sets and on the PX4-RT drone autopilot programs developed by Kopernic team in FR FUI21 CEOS project.