Task Scheduling in the Control Robotics Environment

Abstract

Scheduling problems involved in Control Robotics, a software approach to control engineering are studied. The capability of a multiprocessor system to handle tasks with hard, real-time deadlines is investigated according to whether complete or partial a priori knowledge of the deadlines, computation times and frequencies of occurence of individual tasks is available. A model of preemptive scheduling, the "scheduling game" is introduced to explore mathematical relationships for different scheduling situations. A necessary and sufficient condition for scheduling tasks with simultaneous requests or deadlines is derived. Partial solutions and the difficulties involved in scheduling tasks with distributed requests are discussed. It is shown that in the most general case, there is no globally optimal algorithm in the absense of a priori knowledge about the distribution of requests of future tasks in time.