실시간성과 결함허용을 고려한 임베디드 태스크 스케줄러

Abstract

Embedded systems need to ensure real-time of the task response time depending on the applied fields of it. Nevertheless, existing real-time operating systems in embedded systems perform the task scheduling based on the priority of task without considering aperiodic task. Therefore the properties of both periodic and aperiodic tasks should be considered in order to effectively schedule the real-time tasks. Task fault could be caused by various reasons in real-time system, and real-time system is required to be predictable performance despite occurrence of the fault. In this paper, we propose a task scheduler that considers real-time and fault tolerance in embedded system with a single processor. The scheduler consists of an ECM(executability check manager), BTM(backup time manager), STM(surplus time manager), PTS(periodic task scheduler), FTM(fault tolerant manager) and APTS(aperiodic task manager). The ECM examines the feasibility of a set of periodic tasks before scheduling it and then can reducing the overhead of reconstructing a poorly designed task set. The BTM calculates and manages backup times which is used to recover a faulty task. The STM maintains the processor idle times for aperiodic tasks that can be obtained by excluding from the total processor times to the periodic tasks's execution times. The PTS schedules a set of periodic tasks and supports the execution of aperiodic tasks. The FTM requests back up times to the BTM when a transient fault of task is occurred and recovers a faulty task by restarting it using back-up times. The APTS requests surplus times to the STM when aperiodic tasks generated and schedules aperiodic tasks by using surplus times. We analyze and evaluate if our task scheduler can meets periodic tasks's deadlines and guarantees of aperiodic tasks's completion. And we propose an important level of periodic tasks that can controls response time of periodic and aperiodic tasks. If an important level is raised, the response time of aperiodic tasks increases and response time of periodic tasks decreases. If you want to reduce the response time of aperiodic tasks, it can be obtained by reducing the important level of periodic tasks. Therefore this important level can be used when we formulates task scheduling policy considering the response time of tasks. In order to provide fault tolerance, the scheduler manages backup times and reexecutes a fault task to restore it's conditions. And we analyse and evaluate the proposed fault-tolerance techniques by showing if it can meets deadlines of normal periodic tasks and a faulty task.