Adjustable deadline assignment for end-to-end tasks

Abstract

A task in a multiprocessor system needs to be executed on different processors for the different parts. The different execution parts of the task must be finished by its end-to-end deadline and there are precedence constraints among the parts. The multiprocessor system is usually heterogeneous, which means that the system consists of one general purpose processor and one or more specific purpose computing components, such as digital signal processors. Within considering the different features between the general purpose processor and specific purpose one, we should adopt distinct and suitable schedulability test conditions to ensure the real-time requirements. In this paper, we propose the adjustable deadline assignment (ADA) strategy for the periodic end-to-end tasks in the heterogeneous multiprocessor system. The proposed strategy uses a user-defined parameter to assign sub-deadline for the individual subtasks of each task to improve the successful ratio to pass the scedulability test. Besides, under the ADA strategy, more sporadic jobs can be accepted by the system due to the bias load balance between the processors.