Multiprocessor scheduling of elastic tasks

Abstract

The elastic task model enables the adaptation of recurrent real-time tasks under uncertain or potentially overloaded conditions. The model was originally defined for sequential tasks executing upon a preemptive uniprocessor platform; it was later extended to include tasks with internal parallelism executing on multiple processors. This paper bridges a gap in the theory of elastic task scheduling by considering the multiprocessor scheduling of sequential tasks (i.e., tasks with no internal parallelism). We define algorithms for scheduling sequential elastic tasks under the global and partitioned paradigms of multiprocessor scheduling, and provide a simulation-based comparison of the different approaches.