Improved Priority Assignment for the Abort-and-Restart (AR) Model

H. C. Wong, A. Burns

Research output: Chapter in Book/Report/Conference proceedingConference contribution


This paper addresses the scheduling of systems that implement the abort and restart (AR) model. The AR model requires that preempted tasks are aborted. As a result high priority tasks run quickly and shared resources need not be protected (as tasks only work on copies of these resources). However there is significant wastage as low priority tasks may be subject to a series of aborts. We show that exact analysis of the AR model is intractable. A sufficient but tractable test is developed and is used to address the priority assignment issue. Again an optimal tractable algorithm is not available. The paper develops a priority assignment heuristic that is demonstrated to perform better than existing schemes.
Original languageUndefined/Unknown
Title of host publication7th Junior Researcher Workshop on Real-Time Computing (JRWRTC)
Publication statusPublished - 2013

Cite this