Real-time periodic task scheduling considering load-balance in multiprocessor environment

Abstract

As distributed real-time control system is widely used, multiprocessor real-time tasks scheduling becomes the most important research area. Scheduling of real-time tasks in multiprocessor environment consists of task allocation stage (scheduling in task level) and task scheduling stage (scheduling in process level). A scheduling algorithm considering load-balance of real-time periodic tasks in multiprocessor environment is proposed in this paper. In task allocation stage, the tasks are sorted by the ascending order of their loads firstly. Before each time allocation, we sort the available processors by the descending order of their current loads and then through a mapping strategy assign tasks in turn to corresponding available processors in order to make the loads assigned to each processor balanced as much as possible. In task scheduling stage, we take EDF (Earliest-Deadline First) algorithm to schedule tasks in single processor. The simulation results show that the algorithm proposed in this paper has good and stable load-balance performance.