MC-Fluid: Multi-Core Fluid-Based Mixed-Criticality Schedulingopen access
- Authors
- Lee, Jaewoo; Ramanathan, Saravanan; Phan., Kieu-My; Easwaran, Arvind; Shin, Insik; Lee, Insup
- Issue Date
- Apr-2018
- Publisher
- IEEE COMPUTER SOC
- Keywords
- Real time systems; mixed criticality systems; multiprocessor systems; scheduling algorithm; schedulability analysis; processor speedup factor; fluid scheduling
- Citation
- IEEE TRANSACTIONS ON COMPUTERS, v.67, no.4, pp 469 - 483
- Pages
- 15
- Journal Title
- IEEE TRANSACTIONS ON COMPUTERS
- Volume
- 67
- Number
- 4
- Start Page
- 469
- End Page
- 483
- URI
- https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/60629
- DOI
- 10.1109/TC.2017.2759765
- ISSN
- 0018-9340
1557-9956
- Abstract
- Owing to growing complexity and scale, safety-critical real-time systems are generally designed using the concept of mixed-criticality, wherein applications with different criticality or importance levels are hosted on the same hardware platform. To guarantee non-interference between these applications, the hardware resources, in particular the processor, are statically partitioned among them. To overcome the inefficiencies in resource utilization of such a static scheme, the concept of mixed-criticality real-time scheduling has emerged as a promising solution. Although there are several studies on such scheduling strategies for uniprocessor platforms, the problem of efficient scheduling for the multiprocessor case has largely remained open. In this work, we design a fluid-model based mixed-criticality scheduling algorithm for multiprocessors, in which multiple tasks are allowed to execute on the same processor simultaneously. We derive an exact schedulability test for this algorithm, and also present an optimal strategy for assigning the fractional execution rates to tasks. Since fluid-model based scheduling is not implementable on real hardware, we also present a transformation algorithm from fluid-schedule to a non-fluid one. We also show through experimental evaluation that the designed algorithms outperform existing scheduling algorithms in terms of their ability to schedule a variety of task systems.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Business & Economics > Department of Industrial Security > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.