Integrated Design and Performance Steering of Real-Time Systems

Michigan State University, East Lansing MI

Investigators

Abstract

Distributed real-time systems are often very complex. They are composed of heterogeneous systems intended to perform tasks subject to many constraints, including time constraints in the face of dynamic environmental changes. In order to meet these challenges, a unifying, extensible, compiler-based framework is proposed for the design and engineering of complex real-time systems, with the following research objectives: - To devise a systems engineering framework that includes a systems specification language and multiple problem-solving approaches, as well as an automated mapping between them. - To discover common properties of extant on-line performance visualization tools for parallel and distributed, heterogeneous systems, as well as gather desired properties of the on-line performance analysis and visualization as a real-time middleware. Using the findings, to devise a comprehensive performance and analysis framework as a basis for advanced extensions. - To integrate the systems engineering and on-line performance and analysis frameworks in a way that facilities on-line system steering/reconfiguration of complex real-time systems. The integrated approach should be semi-automated, and allow algorithmic and visual analyses of instrumentation data received from the target system to utilize problem-solving tools.