Job-shifting: An algorithm for online admission of non-preemptive aperiodic tasks in safety critical systems

Modern safety critical systems require certification in order to guarantee correct operation before system deployment. The certification process requires rigorous verification and validation, the efforts for which can be greatly reduced by using resource partitioning. However, Lackorzyński et al. de...

Full description

Saved in:
Bibliographic Details
Published inJournal of systems architecture Vol. 85-86; pp. 14 - 27
Main Authors Syed, Ali, Pérez, Daniel Gracia, Fohler, Gerhard
Format Journal Article
LanguageEnglish
Published Elsevier B.V 01.05.2018
Subjects
Hierarchic scheduling
Mixed-criticality
Online aperiodic admission
Real-time systems
Scheduling algorithms
Mixed-criticality
Hierarchic scheduling
Real-time systems
Scheduling algorithms
Online aperiodic admission
Online AccessGet full text

Cover

More Information
Summary:Modern safety critical systems require certification in order to guarantee correct operation before system deployment. The certification process requires rigorous verification and validation, the efforts for which can be greatly reduced by using resource partitioning. However, Lackorzyński et al. demonstrated that bandwidth reservation for event-triggered (ET) activities in partitioned systems may lead to significant bandwidth loss. In contrast, the online admission of ET activities can prevent bandwidth losses. However, the state-of-the-art approaches for online admission of ET activities fail to fulfill the requirements of safety critical systems as they do not support (i) partitioning, (ii) the industrial mixed-criticality task model or (iii) non-preemptive task execution. In this paper, we present job-shifting algorithm for online admission of non-preemptive aperiodic tasks in partitioned time-triggered environment. Our approach circumvents the bandwidth loss issue with partitioning, and provides guarantees similar to the bandwidth reservation technique such that the certification process of safety critical systems need not be modified. Our approach can be implemented on top of variety of hypervisors and can provide lower response-times for aperiodic tasks. Through evaluation, we demonstrate that our approach efficiently utilizes processor bandwidth and only incurs small scheduling overheads.
ISSN:1383-7621
1873-6165
DOI:10.1016/j.sysarc.2018.01.005