Transactions on Engineering and Computer Science
Generalized Models of Mixed-Criticality Systems for Real-Time Scheduling
Junghwan Lee and Myungjun Kim
Abstract
In this article, we address the divergence theory and practice related to mixed-criticality systems (MCSs) for real-time scheduling from an industry perspective. We identify the practical problems in consideration of design aspects, such as time partitions, the decomposition of tasks, degradation scenarios, fault handling for safety, and criticality mode of MCS. The identified problems in design will induce theoretical limitations and difficulties for adopting prior works to practice. In prior works, criticality modes were viewed as equivalent to the assurance levels; however, these assumptions seem to be too simplified for studies far from practice. They may lead to solutions for real-time scheduling in MCS going in different directions that are highly divergent from the real-world situation. Hence, we propose new models for MCS that are closer to practice. Here, criticality modes are separate from the assurance levels of tasks since the criticality modes exist for degradation or safety scenarios, whereas the assurance levels of tasks, hardware, and software components exist to represent reliability for requirements, designs, verifications, and validations. Based on the newly proposed system models, we will show a way of extending prior works with the proposed mode-based priority protocol for real-time MCSs. The proposed mode-based protocol is easily implanted with existing real-time schedulers or the Automotive Open System Architecture platform.
Keywords: Mixed-Criticality systems; Real-Time systems; AUTOSAR; Safety
Journal Information
- Journal Name: Transactions on Engineering and Computer Science
- Journal Short Name: Trans Eng Comput Sci
- Language: English
- Format of Publication: Online
- Starting Year: 2020
- Focus Subject: Engineering and Computer Science
- Publication Model: Online
- Frequency of Publication: Two Issue a Year
- Review Process: Single blind peer-review by referees
- Time to 1st Decision: 2 to 3 weeks from date of submission
- Time to Acceptance: 4 to 6 weeks, depending upon the required revision cycles
- Time to Publication: 1 to 3 weeks from date of final submission in the forthcoming issue
- DOI Prefix: 10.7620
Indexed and Abstracted In
