Adaptive partitioning of real-time tasks on multiple processors

This paper presents a new algorithm for scheduling real-time tasks on multiprocessor/multicore systems. This new algorithm is based on combining EDF scheduling with a migration strategy that moves tasks only when needed. It has been evaluated through an extensive set of simulations that showed good performance when compared with global or partitioned EDF: a worst-case utilisation bound similar to partitioned EDF for hard real-time tasks, and a tardiness bound similar to global EDF for soft real-time tasks. Therefore, the proposed scheduler is effective for dealing with both soft and hard real-time workloads.