Adaptable Scheduling Framework for Window-Based Real-Time Constraints

This paper analyses the relationships of two types of window-based real-time constraint, the sliding and fixed window constraint. First, based on the definition of "strict harder", the paper gives out the computing system of window constraints. These analyses of the constraints are useful for designing the scheduling algorithms. Thereby, a scheduling framework that is capable of adapting to the changing environment is presented next, it bases on the idea that trying to guarantee unsolvable constraints is not as good as adjusting them. The scheduling framework modulates real-time system by adjusting the constraints on tasks so as to fit the changed resources. Last, evaluation results are presented to show that the scheduling framework is efficient and can largely decrease the mean rate of lost constraints for real-time tasks.