Sporadic task scheduling with overload conditions

Abstract

Hard real-time task scheduling in a dynamic environment has been an important area of research, posing difficult problems. In an overloaded system where periodic and sporadic tasks have computational demands that are greater than the CPU time in that interval, the scheduler faces the question of which tasks must really make their deadlines. Assuming that periodic tasks have priority over sporadic ones, we end up with a system where some sporadic tasks may not make their deadlines. It is known that through the assignment of priorities to tasks based on the earliest deadline policy, there is no way to predict which sporadic task will miss the deadline and which will not. In order to prevent important sporadic tasks from missing their deadlines, we assign each task an importance function that is used by the scheduling algorithm. Generally, the summation of important function values must be maximized to allow the most important tasks to meet their timing constraints. We present two novel scheduling algorithms that try to maximize this summation. We show that these algorithms have better performance compared to related algorithms regarding complexity and benefit optimization.