Ongoing Activities

Since 2022, my research has focused on reliable continuous engineering of Cyber-Physical Systems (CPS), with a strong emphasis on bridging design-time models and runtime operation. Building on previous work on logical time and concurrency semantics, I now develop methods that combine formal modeling, verification and runtime data exploitation for Digital Twins.

A first major axis concerns model hybridization for trustworthy Digital Twins: combining interpretable deductive models with accurate but opaque inductive models (ML/AI). This includes explicit interfaces for heterogeneous models and fusion operators that account for uncertainty, so that confidence-aware decisions can be made during operation.

A second axis addresses design-operation continuity: models are treated as evolving artifacts, continuously updated from execution traces. In this context, we introduced the notion of Influence to characterize how components and environmental conditions impact system-level requirements, and we developed a DSL to support both static risk analysis and runtime diagnosis.

Two additional transversal directions are (i) spatio-temporal specification and verification for safety-critical CPS (including Software Defined Vehicles), and (ii) explicit uncertainty management across heterogeneous models. Since March 2024, as head of the Kairos team, I structure these topics into a coherent scientific roadmap for dependable, continuously engineered CPS.

Selected major publications since 2022
[1] J. Holtmann, J. Deantoni, M. Fockel. Early timing analysis based on scenario requirements and platform models. Software and Systems Modeling, 2022.
[2] S. Zschaler, E. Bousse, J. Deantoni, B. Combemale. A Generic Framework for Representing and Analysing Model Concurrency. Software and Systems Modeling, 2023.
[3] J. Deantoni, P. Munoz, C. Gomes, C. Verbrugge, R. Mittal, et al. Quantifying and combining uncertainty for improving the behavior of Digital Twin Systems. Automatisierungstechnik, 2025.
[4] B. Combemale, P. Vicat-Blanc, A. Blouin, H. Bril El Haouzi, J.-M. Bruel, J. Deantoni, et al. Engineering Digital Twins: A Research Roadmap. EDTconf, 2025.
[5] B. da Silva Oliveira, N. Ferry, J. Deantoni. DemIstifyCPS: A Domain-Specific Language for Influence Modeling in Cyber-Physical Systems. MODELSWARD, 2026.

Software

• Timed and Concurrent Operational Semantics

  TCOS is a recent initiative whose goal is to provide a meta language to specify the behavioral semantics of grammar specified in Langium. From the semantic specification, a compiler to different languages are provided (C++, Python, JS).

• TimeSquare

  TimeSquare is developed for almost 10 years for now. It is dedicated to the specification, simulation, compiling and analysis of logical time constraints in CCSL (Clock Constraint Specification Language, a formal declarative language). It is constantly evolving and can be used either as a plain IDE or as an embedded solver. It has been used to represent the behavioral semantics together with the timing performance in many of our publications.

• MoCCML mapping (ECL)

  MoCCML mapping, formerly named ECL (Event Constraint Language) is a lightweight extension of OCL (Object Constraint Language) with the notion of Event and logical constraints. It is used to specify the concurrent semantics of languages so that, for a specific model, the corresponding CCSL formal model can be automatically synthesize.

• MoCCML

  MoCCML (Model of Concurrency and Communication Modeling Language) is a formal meta-language dedicated to the specification of logical constraints between events, based on constraint automata. These constraints can be used conjointly with a CCSL specification for simulation and analysis. It can also be used in ECL to specify the concurrency of a language or in BCOoL to specify how a coordination pattern between different languages behaves.

• BCOoL

  BCOoL (Behavioral Coordination Operator Language) is a meta-language dedicated to the specification of behavioral coordination patterns. Based on event type from some language interfaces, it specifies how the models that conform such languages have to be coordinated. The specification, make at the language level is used to automatically creates an executable coordination for any model conforming the languages used in BCOoL. The resulting heterogeneous models can be simulated in the GEMOC studio or analyzed in TimeSquare (see here for an heterogeneous simulation from a BCOoL specification)

• Cosim20

  CoSim20: An Integrated Development Environment for Accurate and Efficient Distributed Co-Simulations The development of Cyber-Physical Systems involves several disciplines and stakeholders, which use heterogeneous models and formalisms to specify the system and make early validation and verification. In order to understand the behaviour emerging from the heterogeneous models, a collaborative simulation (co-simulation) can be used. To make it happen, the system engineer must define a correct coordination of the different executable models, which can be distributed over different enterprises. This is an important but difficult (and error prone) task that can not be done without information about the behavioral semantics of each model. In this paper, we introduce an integrated development environment which allows 1) to import different executable models (named simulation units), 2) to graphically connect them with rich connectors and 3) to generate a dedicated, accurate and efficient distributed co-simulation. The framework is based on Eclipse EMF for the modeling part and on ∅MQ for the deployment. It is named CoSim20.

Open Positions

• Please follow our LinkedIn webpage or contact me. We regularly have engineering, post doc, internship and PhD funding.
• a proposal list for DS4H research apprenticeships can be found here: https://webusers.i3s.unice.fr/~deantoni/research_apprenticeships/