Research Interests

Models and Languages for real-time systems
I’m interested in languages and models based on the logical time concept and on synchronous-reactive approaches. I’ve worked on the PsyC language, the CCSL specification language and more traditional synchronous languages such as Lustre and Esterel. In my thesis, I proposed two levels of equivalent semantics descriptions for PsyC, which allowed me to define a framework combining Synchronous-Reactive and Logical Execution Time called Synchronous Logical Execution Time (or synchronous LET).
Formal verification for real-time systems
I’m interested in adapting formal verification techniques to reactive systems, particularly technics based on Model-Checking. I’m working on intermediate techniques between timed automata approaches, and more traditional symbolic approaches such as SAT and BDD. In my thesis, I proposed temporal optimization techniques for PsyC in order to use SAT and BDD approaches.
Implementation of real-time systems
Finally, I’m also interested in the implementation of these systems, whether in terms of compilation or scheduling analysis. I was able to participate in the design of a validation tool (in the context of Krono-Safe), verifying implementation artifacts such as the scheduling plan, or the sizing of communications.

university application (qualification CNU): : link

PhD thesis

Subject and Jury

  • Subject: Methodology for the formal verification of temporal properties for real-time safety-critical applications based on logical time.
  • Academic supervisers: Dumitru Potop-Butucaru and Robert De Simone from Centre Inria d’Université Côte d’Azur.
  • Industrial supervisers: Damien Chabrol and Amira Methni from Krono-Safe (now called Asterios Technologies).
  • Reviewers: Reinhard Von Hanxleden from Kiel University and Pierre-Loïc Garoche from École Nationale de l’Aviation Civile.
  • Examiners: Timothy Bourke from Centre Inria de Paris.
  • Slides: link

Abstract

Safety-critical real-time systems have to respect strict timing constraints. Thus, timing constraints must be considered throughout the software development cycle. As exact computation execution time are generally not known during design, logical time provides a way to abstract time constraints and execution from platform-dependent physical time.

In this thesis, we focus on two main formalisms based on logical time:

  • The Synchronous-Reactive approach totally abstracts physical time by discrete time bases on which computations are triggered.
  • The Logical Execution Time approach uses logical time bases to represent not only triggering instants but also the durations of elementary computations.

We start by unifying Synchronous-Reactive and Logical Execution Time approaches. This provides the natural formal framework for defining the semantics of PSYC, an expressive industrial real-time language. We define two formal semantics for PsyC:

  1. a native big-step semantics preserving the logical durations of time intervals defined by structural operational rules and;
  2. a synchronous small-step semantics defined by translation to a Synchronous-Reactive language expanding time interval durations to a succession of atomic transitions.

We show that the two semantics definitions are equivalent. This formalization of the PsyC semantics enables us to define a formal verification methodology for PSYC based on symbolic model-checking. To reduce the state space during model-checking, we also define an optimization technique inspired by timed automata model-checking. Finally, we show how to encode high-level timing requirements into a clock constraint specification language — CCSL — which are then translated to synchronous observers.

Link to the thesis is available below in the publication list

Main publications

Recent Talks