Open presentation day of MSDL research, similar to research/Summer presentation days in
September 2023,
October 2018,
July 2017,
August 2016, June 2016,
August 2009,
August 2008,
August 2006,
August 2005, and
August 2004.
The presentation day is organised to share the research topics developed by MSDL members and colleagues and their latest results.
The keynote is open for all, while sessions starting at 11:30 are intended for MSDL(++) members to enable productive discussions.
Tuesday 25 April 2025, 10:30 - 18:30.
University of Antwerp, Campus Middelheim, Building G, Room M.G.017.
Abstract: Formal verification aims to prove the correctness of system specifications using logic and mathematical proofs. While verification tools have made significant progress, the growing use of domain-specific models and languages introduces a challenge: bridging the gap between domain experts' models and formal reasoning tools. Traditionally, model transformations are used to translate domain-specific models into verification-friendly representations. However, this approach is prone to semantic inconsistencies, requiring complex equivalence proofs that are difficult to maintain. This talk presents an alternative semantic-level approach, developed over the past decade, which eliminates these issues. At its core is a language-agnostic semantic interface that directly links the dynamic semantics of domain-specific languages to a broader range of analysis tools, including interactive execution, debugging, verification, testing, and runtime monitoring, ensuring precise and efficient system analysis.
Abstract: At the Using Multi-* Modelling to Manage Complexity in Systems Engineering NII Shonan Meeting No. 219 at Shonan Village Center, Japan, the foundations of modelling were discussed by a group of researchers with diverse backgrounds: formal methods, modelling language engineering, surrogate modelling, simulation, contract theory, conceptual and ontological modelling, ... This presentation will give an overview of the insights gained during this meeting. In particular, an attempt is made to give a precise definition and above all, to relate, concept such as modelling, abstraction, refinement, accuracy, approximation, and fidelity.
Abstract: Overview of Current Research on Model Management.
Abstract: The purpose of Systems Engineering is to analyse, design, optimize, operate, and evolve complex Cyber-Physical Systems (CPS). This often happens collaboratively and by following complicated workflows. In order to deal with the increasing complexity of CPSs, the domain of Modelling and Simulation allows for system engineers to focus on their specific domain knowledge. Digital Twins (DTs) are simulation models running continually in parallel with a real-world System under Study (SuS) while being fed the same input stimuli as that SuS. They can be used to analyse, optimize and adapt CPSs. Throughout the (currently quite ad-hoc) creation of such Twinning Systems, a plethora of choices impacts the functionality and performance of the realized system, comprised of the actual SuS and its model. This work identifies four stages at which this variability may appear: (A) Properties of Interest in the Problem Space: an initial choice of the exact goal(s) and purpose(s) for the Twinning System highly impacts the end product. (B) (Conceptual) Architecture and Design: when a choice is made in terms of goal(s), it is important to identify the individual system components required to (functionally) ensure the valid behaviour of the Twinning System. Furthermore, these may need to be combined/federated at a higher, conceptual level. (C) Modelling & Simulation: selecting the exact modelling languages and creating the models of the SuS. (D) Deployment: individual tools and frameworks need to be selected, as well as the middleware, the communication protocols etc. It may be required to do an external analysis of the potential deployment solutions. Choices made in one stage influence the solutions and subsequent possible choices (or configurations) in the subsequent stage. To show these stages and their impact, four distinct proof-of-concepts have been created: (1) a simple LEGO Line Following Robot; (2) a pathfinding TurtleBot; (3) the macro-level movement of ships in the Port of Antwerp; and (4) the simple kinematic movement of a 1D vessel.
Abstract: Experiments are a central notion of the scientific method. As more and more experiments are performed on heterogeneous systems to design increasingly complex systems there is a need for (1) a reusable, computer-representable description of experiments and (2) the management of information related to the experiment (including but not limited to data traces, metadata, system information, environmental specification, workflows). I will present our work on the reproducible design of experiments and management of experiment data with the help of logical reasoning.
Abstract: Engineering teams across a multitude of domains are using ontological models to store and share knowledge with consistency being a critical aspect. We present a framework for correcting a subset of inconsistencies in Cross-Domain Knowledge Models when using distributed version control.
Abstract: The data layer of today's model management solutions often is either centralized or Git-based. We point out a number of limitations of current approaches, such as poor replicability, manually configured access control, centralization, hard-coded 'meta-data', and inflexible encodings. We argue for a set of fundamental features / restrictions (most importantly immutability and capability-based security) for decentralized model management systems to adapt, to solve these problems at their root. We distinguish a fundamental core from non-fundamental applications (such as versioning), that can be built on top.
Maintained by Hans Vangheluwe. | Last Modified: 2025/03/24 08:38:00. |