A task scheduling model and method for control-oriented multicore systems

In real-time control systems, multicore microcontrollers have been more and more widely applied. The workload balance and the responding time of ready tasks are most important factors for real-time multicore embedded systems. In this article, we focus on real-time requirement of multicore processors with shared memory to present one multicore task scheduling model and method based on multi-master model, in which each core can manage the task scheduling and resource allocation like a master once tasks are ready, and declines the response time of ready tasks to be scheduled. And then, based on this model, we propose one multi-master multicore scheduling policy and algorithm to reduce scheduling switch count and to achieve better load-balance. In the last, we have implemented a multi-master scheduling prototype to demonstrate that the approach is promising. This experiments show this model can achieve some benefits for real-time performance and load-balance.