Physical-State-Aware Cyber Resource Management in Mixed-Criticality Systems
Abstract
Cyber-physical systems (CPSes) like ships, cars, and planes require not only different levels of assurance, but also close interacti"ons with dynamically-changing physical environments. While the former has been extensively addressed by exploiting the notion of mix"ed-criticality (MC) systems, the latter has not, especially in conjunction with MC systems. Little has been done to exploit the dyna"mic nature of the physical plant as it progresses along its state trajectory by adapting the behavior of the cyber side to match the" current state, dynamics, and performance metrics associated with the physical plant. The rationale behind this project is the signi"ficant benefits that can be gained by explicitly taking the interaction between the physical and cyber subsystems of CPSes into acco"unt when managing the cyber (computing and communication) resources. Motivated by this, we will conduct an in-depth case study, demo""nstrating the importance of capturing current physical states, and then introduce the problem of achieving efficient utilization of"" cyber resources under varying physical states in MC systems. Our overall goal is to construct, test, and build a physical-state-a""ware cyber resource management framework for MC systems. To this end, we will first develop a physical-state-aware MC task model, wh"ich is a generalization of the existing basic MC task model. We will then propose novel slack concepts tailored to the new task mode"l, which enable efficient utilization of computing resources by considering varying physical states and MC systems. Building upon th""ese concepts, we will develop a physical-state-aware dynamic slack management technique, which not only captures how slack values dy""namically change according to physical states at runtime, but also achieves high resource utilization through efficient slack usage."" We will also develop a physical-state-aware adaptive job-dropping algorithm, which focuses on the key parameters ~ sampling period" and consecutive control update misses ~ in determining system stability and control performance and derives a task schedule so as to improve the overall control performance while guaranteeing both system stability and MC timing requirements. We will conduct case" studies and in-depth evaluations relevant to Navy systems, and demonstrate the power and utility of the proposed framework in achie"ving the efficient use of cyber resources as well as significantly improving control performance without compromising system stability and MC timing requirements.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 20, 2018
- Source ID
- N000141812141
Entities
People
- Kang Shin
Organizations
- Board of Regents of the University of Michigan
- Office of Naval Research
- United States Navy