Application area: Automotive
Researcher in charge: Deepak Vedha Raj Sudhakar
Company Supervisor: Dr. Karsten Albers
Academic Supervisor: Prof. Dr. Frank Slomka
Host: INCHRON GmbH
Secondments: TU/e, Imsys AB

Timing plays a crucial role in Cyber-Physical Systems. The decision on the required types and number of processing elements, the hardware architecture and the software architecture are mainly driven by the required performance and response times (Slomka) (T. Bund, 2013). INCHRON has developed innovative methods and tools for the prediction of the (worst-case) timing behavior of distributed embedded systems by means of simulation (chronSIM) and mathematical analysis (chronVAL). The simulation (chronSIM) covers modelling at an abstract level and more detailed code-level, whereas the method for the mathematical analysis (chronVAL) are so far available only for an abstract modelling level. This research will focus on extending the analysis to code-level and to perform integrated code-level and system-level timing analysis of automotive software. The research will also focus on reducing the pessimism of analytical techniques by this integrated approach.

Expected Contributions:

The primary research goal will be to extend the mathematical analysis (chronVAL) to code-level. This research will use an innovative combination of recently developed methods:

  • The real-time calculus as a strong mathematical theory for schedulability analysis with extensions like approximation methods to support practical relevant systems.
  • The event dependency analysis (EDA), covering event generation within functions.
  • The concept of limiting event-spectrum, which is a comprehensive concept for covering all kind of internal and external dependencies.

EDA and the limiting event-spectrum provides some basic methods necessary for the coverage of code-level analysis, the real-time calculus provides the required background. In this research we intend to extend and combine these methods to reach the goal.

The research covers several of the research lines :

  • Multi-Domain Modelling (RL 1) : Tool support in terms of timing for multi-domain models will be provided. This covers especially the aspects of reliability.
  • Resource Constraint Design (RL2) : The timing analysis covers performance aspects and the resulting constraints on resources as well as aspects of stability, settling time and so on.
  • Platform Aware Control Systems (RL3) : The timing analysis provides the necessary methods to allow the integration of performance relevant platform details into the design of control algorithms.
  • Distributed Coordination (RL5) : The timing analysis methods covers multi-processor and multi-core systems.