A Heuristic Task Periods Selection Algorithm for Real-Time Control Systems on a Multi-Core Processor

The performance optimization problem is investigated for discrete-time control systems on a multi-core platform. An integrated approach which considers both control performance and real-time scheduling aspects is applied to allocate optimal periods to controller tasks. A real-time control system is modeled as a set of directed acyclic graphs with weighted edges in this paper. The system allows producer/consumer relationship between tasks, and the data dependence relationships between tasks are uncoupled by attaching harmonic constraints to task periods. The period assignment problem is formulated as an optimization problem, which minimizes the system performance loss index under multi-core schedulability constraints. A heuristic search algorithm is proposed to solve this optimization problem and select periods for real-time tasks scheduled by rate-monotonic scheduling algorithm. Experimental results demonstrate that the proposed heuristic algorithm is capable of finding a high quality local optimal solution with fast computing speed. The proposed method is applicable to online failure recovery and reconfiguration in real-time control systems.