Compositional Framework for Complex Real-time Systems on Multicore Platforms

Abstract

Major Goals: Modern real-time systems are becoming increasingly complex, autonomous, and resource-constrained. A modern vehicular control system, for instance, has as many as 100 microprocessors running simultaneously and controlling thousands of software functions. Further, to enhance flexibility and to reduce cost, weight and power, these systems are increasingly being deployed on shared multicore platforms, instead of isolated physical hosts. Due to these trends, existing analysis techniques - which were developed for much simpler systems - are becoming inadequate, as they cannot scale to this unprecedented level of complexity. To address the above challenge, this project aimed to develop a novel compositional scheduling and analysis framework for safety-critical and mission-critical real-time systems on multicore architectures. Through new resource-aware interfaces for components and interface analysis methods, the framework not only enables efficient and accurate compositional modeling and analysis of large-scale open systems with tight resource constraints, but it also provides a foundation for component-based development with timing guarantees. Besides these theoretical innovations, the project developed tools and concrete system implementations to demonstrate the utility and effectiveness of the proposed framework, as well as to facilitate technology transfer and follow-on research on robust and interoperable real-time systems.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 30, 2014
Accession Number
AD1225938

Entities

People

  • Insup Lee
  • Linh T. Phan
  • Oleg Sokolsky

Organizations

  • University of Pennsylvania

Tags

Fields of Study

  • Computer science
  • Engineering

Readers

  • Distributed Systems and Data Platform Development
  • Software Engineering