Public lecture on "Building correct Cyber-Physical Systems", prof. Susanne Graf

Public lecture on "Building correct Cyber-Physical Systems", prof. Susanne Graf

We would like to invite you to the public lecture by our visiting Professor:

Susanne Graf, Ph.D. in Computer Science, Grenoble Polytechnical Institute (INPG), France.

Title: Building correct Cyber-Physical Systems: Discussion of some current technology and challenges for the future


Today's cyber-physical systems are physical systems (which may include humans, technical devices ...) which are controlled by tight interaction with a computerized autonomous control system. Typical examples are automated factory control systems, robotic systems, software-controlled cars, planes, medical monitoring systems ... the evolution over the last decade is to move from small, possibly weakly interacting controllers (as an example, take a car from the early 80s with a few automated control systems, such as engine control, braking assistance (ABS) and a window-lift control) to extremely complex and heterogeneous distributed systems (e.g. in a modern car which hosts a system of over 100 electronic control units communicating over multiple networks and execute over 1 mio lines of code), and the future evolution is towards autonomously driving cars evolving in an interconnected open and evolving world. Many such systems are (safety) critical: they must guarantee correct and timely functioning (e.g. an engine control that does not guarantee injection at the right moment is  not a useful injection control, or the stabilisation system of an Ariane-5 rocket that would miss a single 10ms cycle would lead to the explosion of the rocket).

Building such complex, heterogeneous systems in such a way as to be able to guarantee their correct functioning despite environmental hazards (including malicious attacks), failure of hardware components etc., is still a very challenging task. There is no unanimously approved approach, no fully integrated tool support, the design practices differ from application domain to application domain, and even from manufacturer to manufacturer ... in a context where the difficulties to be overcome are rapidly increasing (from statically defined control functions to dynamically evolving ones (adaptivity, learning), from once-for-lifetime design to updatable and dynamically evolving systems, from closed-world to open systems (IoT) ... posing many additional challenges.

The presentation will be organized in three parts:

  1) A general motivation and introduction to the topic, taking as a meaningful example for a global design approach the Autosar approach gradually put into practice in the European automotive industry to illustrate the overall complexity and the interconnection between the different problems to be solved.

  2) An overview on a small subset of technologies that are at least should be, typically part of a systematic design approach in almost any application domain where the systems to be built have an important criticality level. For each one, we discuss its strengths and weaknesses, and what it is able to guarantee. We will also explain the challenges imposed by the future evolution of systems to be built and discuss whether and how they can be adapted to these new challenges. The selected issues will include the following ones (a list of other important techniques and issues, but which this talk does not address will be provided)

    a) Verification technology used/useful in the context of cyber-physical systems, typical properties to be enforced, how to ensure their overall validity without verifying the global model (which would have no chance to scale)

    b) Communication networks and protocols for embedded systems

    c) High-level "programming": approaches for software architecture design (and validation) and of function design

3) A question and answer and general discussion session

Lecturer’s bio:

Susanne Graf received her Ph.D. in Computer Science in 1984 from Grenoble Polytechnical Institute (INPG). Currently, she is a “Directeur de Recherche” (Research Professor) at CNRS in the Verimag research laboratory in Grenoble.

 Her research interests include theories, algorithms, and tools for modeling and verification, as well as the application thereof to the design and certification of embedded and real-time as well as mixed-criticality systems. Presently, she is focussing on knowledge and contract-based methods to achieve scalability for multiview models.  Some of her main contributions are in the domain of abstraction and abstract interpretation where in particular, she has proposed together with Hassen Saidi a method called "predicate abstraction".

 She has been the Organisation Chair of ETAPS 2002 in Grenoble, she has been the PC chair of TACAS 2000, SPIN 2004, ATVA 2006, and FORTE 2015. She is on the editorial board of Springer’s STTT journal and has participated in a large number of Program and Steering Committees.

She has participated in and coordinated a large number of National and European projects, as well as collaborations with industry. In particular, she has been the overall coordinator of the European IST project OMEGA on Correct Development of Real-Time Embedded Systems (2002-2005), the coordinator of the sub-project on “Design and Validation” of the European IP SPEEDS on rigorous System design in the embedded domain (2006-2011), and the coordinator of the Modelling &Validation cluster of the European NoE ARTIST on the Design of Embedded Systems (2004-2012).

На главную




Сайт находится в технической разработке