Integer Linear Programming Models for Multicore Cyclic Executive Generation with Task Dependencies

The generation of cyclical executives (CEs) in multicore platforms amplifies challenges previously restricted to unicore systems. In this work, we propose models based on Integer Linear Programming (ILP) for building ECs in multicore environments, explicitly integrating precedence and exclusion relationships between tasks, critical aspects neglected in previous works. The proposed models allow the generation of preemptive and non-preemptive CEs under partitioned and global allocation schemes, ensuring temporal predictability even in scenarios with complex interdependencies. The experimental results demonstrate that the models produce optimal cyclical executives for various sets of tasks, proving their effectiveness in small and medium-scale scenarios.