Research

Our research focuses on the model-based engineering of dependable and intelligent cyber-physical systems. The central objective is to develop systematic engineering methods for specifying, structuring, analyzing and optimizing complex software-intensive systems.

The research focuses on engineering foundations for software and cyber-physical systems, as well as on the analysis, optimization and runtime adaptation of these systems.

Engineering Foundations

Model-Based and Compositional Engineering

Models provide explicit representations of system structure and behavior and form the basis for specification, analysis, transformation and implementation. My early research addressed component-based and model-driven engineering of distributed mechatronic and real-time systems, including behavioral synthesis, model-based testing, formal verification, legacy integration and resource-aware development (A Survey of Approaches for the Visual Model-Driven Development of Next Generation Software-Intensive Systems, 2006; Modeling Techniques for Software-Intensive Systems, 2008).

A central principle is compositionality. System properties are related to models and properties of individual components and their interactions. This supports systematic reasoning about complex and evolving systems, including formal integration of existing components (Combining Compositional Formal Verification and Testing for Correct Legacy Component Integration in Mechatronic UML, 2008) and synthesis of component behavior from scenarios (Synthesis of Timed Behavior from Scenarios in the Fujaba Real-Time Tool Suite, 2009).

This work provides the methodological basis for subsequent research on architectures, contracts, refinement, design-space exploration and adaptation.

Architecture and Contract-Based Systems Engineering

Architecture-centered systems engineering extends component-based methods across abstraction levels and engineering viewpoints. Models represent requirements, functions, logical structures and technical architectures and make relationships between engineering artifacts explicit (Introduction to the SPES Modeling Framework, 2012; Technical Viewpoint, 2012).

Contracts specify assumptions and guarantees of components and subsystems and enable compositional reasoning about their integration and evolution. This includes contract-based analysis of evolving systems (Contracts for Evolving Systems, 2013) and compositional analysis of non-functional properties such as scheduling and timing (Contract-Based Compositional Scheduling Analysis for Evolving Systems, 2013).

Model-based engineering, architectures and contracts constitute the engineering foundation for systematic analysis, optimization and adaptation.

Analysis, Optimization and Adaptation

Formal Refinement and Scalable Verification

Formal refinement establishes relationships between models, architectures and specifications and supports the preservation and reuse of verification results during system evolution. Early work investigated refinement checking for dynamic and real-time systems (Specification and Refinement Checking of Dynamic Systems, 2009; Automata-Based Refinement Checking for Real-Time Systems, 2013) and methods for reducing repeated validation effort (Reducing Re-Validation Efforts for Real-Time Systems, 2015).

Current work extends this concept to formal properties. Refinement relations between temporal properties allow their logical strength and dependencies to be analyzed and can reduce the set of properties required for verification (Property Refinement in Linear Temporal Logic: Formal Semantics and Algorithms for Software Verification, 2026; What Do We Really Need? Identifying Minimal Sufficient Temporal Properties via Refinement, 2026).

The objective is scalable verification based on explicit and reusable relationships between models, specifications, properties and verification results.

Design Space Exploration and Optimization

Cyber-physical systems typically admit multiple architectures, mappings and configurations with different functional and non-functional characteristics. Design Space Exploration provides methods for representing and evaluating these alternatives systematically.

My research combines model-based DSE with constraints and quantitative objectives, including real-time behavior, reliability and security (A Design Space Exploration Framework for Model-Based Software-Intensive Embedded System Development, 2013; Multi-Objective Design Space Exploration for Cyber-Physical Systems Satisfying Hard Real-Time and Reliability Constraints, 2014; Integrating the Security Aspect into Design Space Exploration of Embedded Systems, 2014).

Current work transfers these principles to resource-constrained intelligent systems and runtime optimization, including energy-efficient TinyML (Multi-objective DSE for Energy Efficient TinyML Applications with Real-time Constraints, 2025) and knowledge-driven reduction of configuration spaces (Runtime Self-Optimization Through Knowledge-Driven Design Space Reduction, 2026). The objective is to integrate engineering constraints, system analysis and optimization rather than treating optimization as an independent process.

Runtime Adaptation and Intelligent Systems

Runtime adaptation addresses systems whose configuration or behavior must change in response to their environment, available resources or operational objectives. The methodological basis originates in research on self-optimizing mechatronic systems and dynamically adapting component architectures (Modeling Collaborations with Dynamic Structural Adaptation in Mechatronic UML, 2008; Safe Online-Reconfiguration of Self-Optimizing Mechatronic Systems, 2008; Reusing Dynamic Communication Protocols in Self-Adaptive Embedded Component Architectures, 2011).

Current research combines these principles with digital twins, optimization and intelligent system components. Models and contracts represent admissible system configurations and constraints (Contract-based Digital Twin Synthesis for Autonomous Safety Critical Systems, 2023), while design-space exploration and engineering knowledge support efficient runtime decisions (Runtime Self-Optimization Through Knowledge-Driven Design Space Reduction, 2026).

Artificial intelligence is considered within this engineering context: learning-based components provide capabilities such as perception or prediction, while models, constraints and formal properties provide engineering knowledge for their systematic integration into cyber-physical systems.

Application Domains

The methods are developed and evaluated across application domains including automotive and autonomous transportation systems, railway and avionics systems, industrial automation, networked mobility, UAVs, IoT and smart farming. These domains are application contexts rather than separate research areas and provide concrete systems for evaluating model-based engineering, formal analysis, design-space exploration and adaptation under different functional, real-time, resource and dependability constraints. Recent applications additionally involve embedded AI and TinyML, digital twins and optimization of autonomous and agricultural cyber-physical systems.