Validating Cyber-Physical Systems requires co-simulation of continuous physical dynamics alongside discrete control logic. However, standard fixed-step coupling often limits accuracy at the interface. Discretization errors distort signal behaviour between updates, and asynchronous events, such as safety threshold violations, are frequently detected with critical delays. In this paper we address these limitations by extending the FERAL simulation framework with a derivative-aware communication interface and an adaptive rollback mechanism. Our approach enables subsystems to extrapolate inputs continuously using First-Order Hold and allows the simulator to retroactively correct timing errors by re-simulating critical windows with high precision. We validated this methodology using an Adaptive Emergency Brake case study. Although the standard baseline simulation failed to prevent a collision due to an 891 ms detection delay, our proposed method achieved a safe stop with a response latency of < 1 ms, effectively eliminating the synchronization error. Performance analysis confirms that this 99.8\% accuracy improvement is achieved with only a marginal computational overhead, utilizing the hierarchical architecture of the framework to execute targeted rollbacks of only the affected components to the exact time of the event.

Validating Advanced Driver Assistance Systems (ADAS) requires executing millions of simulation scenarios, which necessitates efficient coupled simulations that integrate heterogeneous models. However, these simulations contain events that are discontinuities or sudden state changes and pose significant challenges. If not detected properly, such events can cause numerical instabilities, erroneous results, and costly rollbacks. This paper presents a component-based event detection framework for hybrid simulations that captures, classifies, and analyzes events during execution. Our approach decomposes simulation scenarios into reusable components, maps raw component outputs to semantic event categories, distinguishes new discontinuities from state continuations, and builds correlation models revealing temporal relationships between event types. Evaluation in an urban traffic scenario involving lane merging, traffic signal coordination, and pedestrian crossing demonstrates that our framework successfully identifies distinct behavioral signatures across components with only ~3% overhead. Analysis reveals that 96% of reported events represent stable behavior suitable for coarser temporal resolution, suggesting 24.5% potential computational savings through adaptive time step scheduling. Our contribution is a systematic event detection infrastructure that enables informed optimization decisions for simulations in CI/CD (Continuous Integration/Continuous Delivery) pipelines, providing a foundation for efficient ADAS validation through virtual testing.

Digitalization plays an important role in improving the sustainability of manufacturing systems. Documenting and sharing the Product Carbon Footprint (PCF) improves traceability along the value chain. However, current standards do not define in detail to what extent digitalization is necessary for calculating the PCF and what information is required. To address this problem, we present a novel approach based on Green Digital Twins (GDT) to determine the product carbon footprint in the manufacturing domain. The data models of the GDT provide a modeling approach to represent the necessary activity data required for PCF calculation. The calculation models provide a practical solution for companies to calculate the PCF based on the digitalization degree of the manufacturing system. The data models are implemented using Asset Administration Shell and submodels. The evaluation demonstrates PCF calculation for batches of products with GDT during the setting process of a drying chamber.

Nowadays, the automation industry is undergoing many changes, manufactures must react to fast changing market demands and more individual customer requirements. Recept-based, Service-Oriented Architectures enable the efficient adaption of production processes to new operating conditions. However, to ensure production performance, service-oriented production should also be complemented by adequate scheduling approaches to guarantee critical performance factors. We present a Digital Twin-based approach that performs integrated process planning and scheduling for service-based production. For our approach, we have identified a common set of input data required for integrated process planning and scheduling. We use Deep-Q-Network, which is a deep Reinforcement Learning method, to derive near optimal schedules for production conditions described in Digital Twins. If an unforseen event happens during the production, our approach is able to adapt current schedules to the changed operating conditions. The case study shows that our approach is able to derive near optimal schedules for customized products and adapt its currents schedules for new orders with different production goals.

Many Industry 4.0 use cases require the integration of live data, e.g., from sensors and devices. However, the large number of legacy fieldbus protocols and proprietary data formats turns this integration into an effort-consuming task. As the number of digital twins in a factory increases rapidly, data source integration has to scale well. Until now, little guidance is available on how to implement this integration in a scalable and reusable manner for Industry 4.0. To close this gap, we define five architecture blueprints based on our experience in various Industry 4.0 projects. These blueprints detail various integration scenarios differentiated by key attributes like frequency of data consumption and data production. In these architecture blueprints, two core components, the Updater and the Delegator, are identified. By providing and evaluating our open-source implementation of these two components, we show the feasibility of the defined blueprints. Utilizing the provided open-source components and the defined architecture blueprints will benefit practitioners as well as researchers when it comes to data integration with digital twins.

The digitization of value chains is an ongoing challenge in production. The Asset Administration Shell (AAS) aims to address this issue. Besides standardization activities, there is however little architectural guidance on how to bridge the gap between potential AAS use-cases and realization.In this paper, we describe four AAS related use-cases that we derived from 15 application projects which adopted the AAS into industrial contexts. For each of the four use-cases, we devise an architecture blueprint that documents our experiences when applying the AAS. By utilizing these blueprints, practitioners can benefit from our experiences when implementing the AAS and bridge the gap between use-cases and implementation more easily.

Trust is a fundamental aspect in enabling self-adaptation of intelligent systems and in paving the way towards a smooth adoption of technological innovations in our societies. While Artificial Intelligence (AI) is capable to uplift the human contribution to our societies while protecting environmental resources, its ethical and technical trust dimensions bring significant challenges for a sustainable self-adaptive evolution in the domain of safety-critical systems. Inspired from the safety assurance case, in this paper we introduce the concept of trust assurance case together with the implementation of its ethical and technical principles directed towards assuring a trustworthy sustainable evolution of safety-critical AI-controlled systems.

The increasing importance of automation and smart capabilities for factories and other industrial systems has led to the concept of Industry 4.0 (I4.0). This concept aims at creating systems that improve the vertical and horizontal integration of production through (i) comprehensive and intelligent automation of industrial processes, (ii) informed and decentralized real-time decision making, and (iii) stringent quality requirements that can be monitored at any time. The I4.0 infrastructure, supported in many cases by robots, sensors, and algorithms, demands highly skilled workers able to continuously monitor the quality of both the items to be produced and the underlying production processes. While the first attempts to develop smart factories and enhance the digital transformation of companies are under way, we need adequate methods to support the identification and specification of quality attributes that are relevant to I4.0 systems. Our main contribution is to provide a refined version of the ISO 25010 quality model specifically tailored to those qualities demanded by I4.0 needs. This model aims to provide actionable support for I4.0 software engineers that are concerned with quality issues. We developed our model based on an exhaustive analysis of similar proposals using the design science method as well as expertise from seasoned engineers in the domain. We further evaluate our model by applying it to two important I4.0 reference architectures further clarifying its application.

Abstract Traditionally, the quality of a software or system architecture has been evaluated in the early stages of the development process using architecture quality evaluation methods. Emergent approaches like Industry 4.0 require continuous monitoring of both run‐time and development‐time quality properties, in contrast to traditional systems where quality is evaluated at specific milestones using techniques such as project reviews. Considering the dynamics and minimum down‐time imposed by the industrial production domain, it must also be ensured that Industry 4.0 system evaluations are continuously performed with high confidence and with as much automation as possible, using simulations, for instance. In this regard, there is a need to develop new methods for continuously monitoring and evaluating the quality properties of software‐based systems for Industry 4.0, which must be supported by automated quality evaluation techniques. In this research we analyze traditional architecture evaluation methods and Industry 4.0 scenarios, and propose an approach based on Digital Twins and simulations to continuously evaluate runtime quality aspects of the architecture and systems of industrial production plants. The evaluation is based on the instantiation of our approach for a concrete demand of an automation plant in the automotive domain.

Art. 106562

In order to enable changeable production of Industry 4.0 applications, a production system should respond to unpredictable changes quickly and adequately. This requires process planning to be performed based on the real time operating conditions and dynamic changes to be handled with cognitive skills. To meet this demand, we present a process planning approach using digital twins and reinforcement learning to derive near-optimal process plans. The digital twins enable access to real-time information about the production system. They also constitute the environment for training the agent of the reinforcement learning method. The environment works as a virtual plant, containing the attributes of the product and resources, and uses simulation models of the resources to calculate the reward for an action in terms of reinforcement learning. Reinforcement learning enables our approach to derive process plans via trial and error. Besides the virtual plant, our approach has a planner, which plays the role of the agent to derive near-optimal plans by trying different actions in the virtual plant, and observes the rewards. We apply the Q-learning algorithm to derive near optimal process plans. The evaluation results show that our approach is able to derive near-optimal process plans for different problem sizes. The evaluation also demonstrated the planner's ability to identify by itself which action to take in which situation. Consequently, no modeling of the preconditions and effects of the actions is necessary.

Temporal decoupling is a state-of-the-art method to speed up virtual prototypes. In this technique, a process is allowed to run ahead of simulation time for a specific interval called quantum. By using this method, the number of synchronization points, i.e. context switches, in the simulator is reduced and therefore, the simulation speed can be increased significantly. However, using this approach can introduce functional simulation errors due to missed synchronization events. Thus, using temporal decoupling implies a trade-off between speed and accuracy and the size of the quantum must be chosen wisely with respect to the simulated application. In loosely timed simulations most of the functional errors are tolerable for the sake of simulation speed. However, for instance safety critical errors are rare but can lead to fatal results and must be handled carefully. Prior works present mechanisms based on checkpoints (storing/restoring the internal state of the simulation model) in order to rollback in simulation time and correct the occurred errors by forcing synchronization. However, checkpointing approaches are intrusive and require changes to both the source code of all the used simulation models and the kernel of the simulator. In this paper we present a non-intrusive rollback approach for error-free temporal decoupling, which allows the usage of closed source models by using Unix’s fork() system call. Furthermore, we provide a case study based on the IEEE simulation standard SystemC.

The growing usage of Domain Specific Modeling Languages (DSML) for architecture view frameworks induces a need for automatic verification of non-functional model properties like completeness and consistency. However, we argue that the high demand for tailored architecture view frameworks is not complemented by appropriate constraint specification facilities. OCL is a common language for defining modeling constraints, but industry user reports indicate that despite its accuracy, it is too complex to be adopted in industrial scale. Approaches that were proposed to simplify the use of OCL either operate on technical formalisms or lack tool support to express new, or more complex types of constraints that can be validated automatically on the model. To address this challenge, we present a constraint modeling framework for the specification and validation of constraints on DSMLs. A Constraint Modeling Language (CML) created based on this framework provides a high level constraint specification en- vironment by using extensible template implementations to enable the automatic validation in computer aided software engineering (CASE) tools. We evaluate the approach in different industry projects and observe that using the proposed framework enhances understandability and effectiveness of constraint specification.

In this paper, we present experiences and the outcomes of applying domain-specific modeling techniques to the domain of diagnostic systems within an industry project. The purpose of the project was to develop a diagnostic system that is easy to configure to different facilities without a deeper knowledge of software development. Although the project was relatively small, we will show that the use of domain-specific technologies proved to be of value, was superior to traditional approaches, and turned out to be a key enabling technology in the project.

In this paper, we present black burst synchronization (BBS), a novel protocol for decentralized network-wide tick synchronization in wireless ad hoc networks, located at MAC level. We argue that tick synchronization is sufficient for multi- hop medium slotting, and that it can be used as a basis for time synchronization. BBS is based on the exchange of synchronized tick frames, which are protected against collisions by a special encoding with black bursts. We deduce a deterministic bound for tick accuracy that depends linearly on maximum network diameter, clear channel assessment delay jitter, and transceiver switching time only. In an improved version of BBS, tick accuracy becomes independent of switching time. We show that BBS has low complexity in terms of communication, computation, storage, and energy consumption, provides a low and deterministic convergence delay, and is robust against node movements and node failures. We have implemented BBS on MICAz motes, and have experimentally validated its predicted behavior. In addition, we have simulated BBS to study its behavior in larger and denser networks.

In this paper, we present a new MAC layer called MacZ, specifically devised for flexible QoS support in mobile ad- hoc networks. MacZ provides multi-hop tick synchronization, network-wide medium slotting, contention-free transmission with global reservations, contention-based transmission with priorities, and is robust against topology changes. We describe the main functionalities of MacZ, show results of performance simulations that provide evidence for the effectiveness of the QoS mechanisms, and discuss related MAC layers.