Autonomy and evolution in safety-critical systems -Mission
impossible, necessary evil, or great opportunity?
Embedded (computing) systems have become a key innovation driver for industry, thereby reaching an innovation speed even in traditional products that was unthinkable a decade ago. The flexibility of embedded software and the configurability of embedded systems networks enable continuous change after product deployment. Software updates of vehicles in the shop are becoming as common as software updates of smart phones. However, while PC or smart phone updates are often automatic and incremental, software updates of cars or other complex and safety critical systems are thoroughly lab-tested using models and prototypes under controlled conditions before they are released to the field.
The new research group CCC studies continuous change in safety-critical and high-availability systems without the need for expensive and time-consuming lab integration tests, even for open systems. Eight scientists from Electrical Engineering and Computer Science, together with their students, combine principles of contracting, self-protection, monitoring, self-optimization with new methods of timing and safety analysis in order to achieve a coherent approach to this engineering challenge. CCC will therefore look for ways to control the variety of updates of different software applications in order to anticipate side effects, and to make computer platforms more robust to ensure proper functioning of the various software subsystems.
The research program addresses new methods to develop and control ESPs which efficiently and robustly integrate several changing applications under high requirements to real-time, safety, availability, and security. Research shall cover network and computing nodes including hardware and run-time environment (RTE), i.e. all the software needed to run applications, such as basic software, operating system(s), communication stack and application program interface (API).
The following video explains the CCC research approach.