Dependable Computing and Fault Tolerance

HEAD

PHD DELEGATE

Our focus is on large and networked computing systems comprising a variety of components (supercomputers to embedded systems) in which software and services play a decisive role. It is important to take into account the evolution of such systems (functional, environmental, technological) involving new threats and classes of faults. Our research addresses four major challenges: mobility, evolvability and autonomy, openness, and reactivity. A key feature is the breadth of faults considered: accidental faults (physical, software) and malicious interaction faults. Our approach combines analytical and experimental methods to meet these challenges.

Architecture (Fault prevention and Fault tolerance)

Fault tolerance software for dynamic distributed systems

The challenge is to tolerate faults despite sources of uncertainty such as the locality of knowledge (no global view) and the asynchrony of distributed nodes.

Adaptive fault tolerance

Dynamic adaptation of fault tolerance mechanisms is a means of ensuring the persistence of dependability in the face of changes in the system or its environment.

Safety monitors for critical autonomous systems

This work focuses on autonomous robots and vehicles operating in spaces shared with humans. They implement monitoring mechanisms that can trigger safety interventions when a dangerous situation is detected.

Security architectures

The aim is to design architectures that are resistant to attacks and malicious acts.

Model-based intrusion detection

Our work focuses on intrusion detection in systems for which there are no existing attack bases: critical embedded systems and the Internet of Things. In these cases, intrusions cannot be detected by searching for known attack signatures, and a model-based approach is particularly relevant.

Personal data protection

The challenge is to enable the use of digital applications and services while protecting personal data.

On-line learning-based anomaly detection

This work focuses on the early detection of anomalies affecting services deployed in the cloud.

Multi-core architectures and mission-critical systems

Multi-core architectures introduce a lack of predictability in execution times. Nevertheless, their introduction in mission-critical systems will become essential to meet the growing need for embedded computing power. Our work aims to offer solutions to this problem.

Analysis (Fault removal and Fault forecasting)

Abstractions and models of connected dynamic systems

This work focuses on graphs as models of interaction structures, from an algorithmic point of view.

Experimental characterization of social interactions

This work exploits location data from human interaction, for example to deduce mobility models or to study collective decision-making processes.

Software testing of stand-alone systems

The focus is on simulation-based testing methods. The system is immersed in virtual worlds, where it is confronted with a wide range of test situations.