Domain #6: Safety

Home > Areas of competence > Safety

Modelling and demonstrating the dependability of systems

The massive combination of heterogeneous components (hardware, software, human) induces strong constraints, and then increases the system’s complexity. It occurs on the various dimensions of dependability: reliability, diagnosis, availability, maintainability, safety and security.
Whilst these constraints are correctly understood and used for some industrial sectors, current dependability approaches nevertheless have shortcomings.
On the one hand, the coupling of safety and security approaches is weak. On the other hand, and in particular for hardware systems, they struggle to consider their dynamics, their structures or their heterogeneities.

Challenges

The current challenges are shared by various industrial sectors.. The increasing complexity of systems, as well as the different dimensions of dependability must be assessed with new innovative methods and tools.

Positioning of the institute

Dependability is the focus of many of IRT SystemX’s R&D projects, particularly in the field of autonomous mobility and artificial intelligence. In this context, the institute provides state of the art solutions and carries out more upstream research works addressing three main challenges: how to handle heterogeneity and non-stationarity of systems? How to assess dependability including formal approaches? How to improve the metrology of dependability studies?

Roadmap

Scientific and technological challenges	Related research fields
Safety analysis of non-stationary and heterogeneous systems	• Systems of systems and autonomous systems • Cyber-physical systems (asynchronous/synchronous, real-time, embedded, local or distributed) • Security protocol
Methods and tools for assessing the dependability	• Combining cyber security and safety • Smart validation of autonomous systems • Verification (model checking, test case generation, deductive verification) • Behavioural modelling
Metrology of the quality of dependability studies	• Homologation/certification • Automation of proposals for solutions to optimise dependability (redundancy, reliability improvement, maintenance strategies, etc.) • Interpretability and reliability of results in the field of artificial intelligence • Models consistency / synchronisation