Job-shop Scheduling Over a Heterogeneous Platform

Abstract

Real-time scheduling involves determining the allocation of platform resources in such a way tasks can meet their temporal restrictions. This work focuses on job-shop tasks model in which a task have a finite number of nonpreemptive different instances (jobs) that share a unique hard deadline and their time requirements are known until task arrival. Non-preemptive scheduling is considered because this characteristic is widely used in industry. Besides job-shop scheduling has direct impacts on the production efficiency and costs of manufacturing systems. So that the development of analysis for tasks with these characteristics is necessary. The aim of this work is to propose an online scheduling test able to guarantee the execution of a new arriving task, which is generated by human interaction with an embedded system, otherwise to discart it. An extension of the schedulability test proposed by Baruah in 2006 for non-preemptive periodic tasks over an identical platform is presented in this paper. Such extension is applied to non-preemptive tasks that have hard deadlines over a heterogeneous platform. To do that, some virtual changes over both the task set and the platform are effectuated.