Analyzing Real-Time Scheduling of Cyber-Physical Resilience
Abstract
Cyber-Physical Systems (CPS) involve software executing on a computer that interacts with its physical environment. Common steps in the design and analysis of such systems are: model the physical environment, develop software to interact with this physical environment, specify timing requirements of software, configure the software (e.g., assign priorities), and then analyze the timing requirements for a given configuration. This approach works but tends to have low resilience to disruption. With the pervasive use of CPS, there is an increasing need for developing timing analysis methods that achieve increased resilience by modeling the linkage between the execution of software and the physical environment. In this paper, we present a new model that describes the current state of the physical environment in terms of how tolerant it is to disruption of the software system; we call this model Cyber-Physical Resilience (CPR). We present an exact schedulability test for this model and implement a tool that performs this schedulability test. Through evaluation of randomly-generated tasksets, we find that (i) for tasksets with at most five tasks, for all tasksets in our evaluation, our new schedulability test never took longer than 15h, (ii) in most cases, our new schedulability test finishes much faster (seconds/minutes),and (iii) thanks to our CPR model, our new schedulability test makes it possible to guarantee schedulability on a single processor even for tasksets with utilization 400%.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2018
- Accession Number
- AD1087870
Entities
People
- Bjoern Andersson
- Dionisio de Niz
- Sagar Chaki
Organizations
- Carnegie Mellon University