In this paper, the use of Large Language Models (LLMs) for the configuration of hybrid TSN/5G networks is investigated. We discuss promising use cases where LLMs offer significant potential to simplify complex configuration tasks. Particularly, we consider two important scenarios: In the first scenario, the LLM functions as an engineering component enhancing traditional network control entities such as the Centralized User Configurations (CUCs) and Centralized Network Configurations (CNCs) of TSN networks and interacting with the 5G control plane. In the second scenario, the LLM serves as an interactive assistance tool for users performing manual configuration tasks. For these use cases, an LLM-based architecture for network configuration is proposed, which consists of a Retrieval-Augmented Generation (RAG) system, a verification component, and an orchestration layer. Within the framework of this architecture, we introduce LLM-based methods to enhance the reliability of configuration in complex real-time networks, leveraging strategies such as divide and conquer, prompt engineering, and verification.

The deployment of autonomous mobile robots in Industry 4.0 environments requires reliable and adaptable SLAM (Simultaneous Localization and Mapping) techniques. This paper presents a comparative evaluation of multiple SLAM algorithms that take advantage of both LiDAR and visual inputs on a Clearpath Jackal robot using ROS2 Humble. Unlike existing studies limited to sensor-specific SLAM approaches, our work systematically benchmarks algorithm performance in a testbed at SmartFactoryOWL, Lemgo, under varying sensor configurations and environmental conditions to assess their applicability in real-world industrial scenarios. We evaluated traditional algorithms such as GMapping, Hector, and Cartographer, along with Visual SLAM frameworks such as ORB-SLAM3 and RTAB-Map. Experiments are conducted in both Gazebo simulation and real-world environments modeled on SmartFactoryOWL. Key performance indicators including mapping accuracy, localization accuracy, robustness to dynamic changes, and real-time performance are systematically analyzed. This study aims to identify suitable SLAM solutions tailored for smart factory applications fostering better human-robot collaboration, thereby contributing towards robust indoor autonomy for mobile robot navigation.

5G (and beyond) cellular networks have gained much attraction in the past for various application domains in context of industry automation, autonomous and cooperative mobility, as well as smart cities. A main commonality of those use cases is their strong reliance on specific Quality-of-Service (QoS) metrics, which are crucial for seamless operation of those applications. This especially holds true for 5 G use cases in the aviation domain, where a prominent application is the Virtual Table Inspection (VTI) - a process in which aircraft maintenance, disassembly, and reassembly are continuously monitored in real time via high-quality video and audio streams, enabling end-to-end tracking of quality control for all aircraft components. Providing concurrent high-quality video streams requiring high data rates and low latencies together with guaranteed packet delivery ratios is a main challenge of 5 G networks - that particularly holds for 5 G campus networks which are often difficult to scale. Our project DeSiRe-NG tackles exactly these shortcomings by providing a versatile architecture to detect under-performance of 5G networks. Further, our proposed architecture allows to generate the Digital Twin (DT) of a 5G network’s QoS, which helps to improve the research and development process of novel networking architectures and applications, as well as aid the continuous supervision of a network. In a first proof of concept study, we present the functionality of our system in an industrial environment.

With the advent of mmWave communication in next-generation wireless networks and their associated non-line-of-sight coverage constraints, Integrated Access and Backhaul (IAB) technology has become increasingly relevant. Current IAB approaches assume fixed power connections, while the next step towards total mobility involves battery-operated IAB nodes. The potential scalability benefits of nomadic 5G+ networks with fully wireless nodes can be particularly valuable in dynamic application scenarios, offering the required adaptability while satisfying stringent network utility constraints. However, managing the balance between operational costs, pertaining energy efficiency, as well as communication integrity in IAB networks is a significant challenge and an active research topic. Consequently, this work in progress paper explores recent advances in IAB network management that aim to enhance energy efficiency and network utility. It identifies new research opportunities and discusses their challenges. Additionally, a conceptual Zero-Touch Management (ZTM) framework for optimizing the operational efficiency of IAB networks is proposed. The framework considers dynamic link scheduling and resource allocation strategies for improving energy usage. This ongoing study intends to provide preliminary insights into the current state of research on IAB network management, emphasizing the topics of energy efficiency and service quality as future research directions.

Mobile robots are essential in various industries, with Simultaneous Localization and Mapping (SLAM) technology playing a crucial role in their autonomy. This work-in-progress paper lays the foundation for evaluating 2D LiDAR-based SLAM algorithms for implementation on a Clearpath Jackal robot in a smart factory environment. The study focuses on three SLAM algorithms: GMapping, Cartographer, and Hector SLAM. A real-world smart factory, serving as a case-study location, is modelled in the Gazebo simulator to evaluate the selected algorithms according to mapping quality, location accuracy, and performance consistency. The simulation uses a hardware-in-the-loop approach, where LiDAR data is processed by the physical Jackal robot, ensuring realistic testing conditions. The findings from the simulation, including the key factors influencing the performance metrics, are validated through real-world testing. This paper outlines the methodology for both simulation and real-world deployment, setting the stage for determining the most suitable SLAM algorithm for efficient and accurate mapping and localization within the operational constraints and requirements of a smart factory environment. Additionally, preliminary insights into factors affecting SLAM performance in the real-world and the relative strengths and weaknesses of each framework are discussed.

Current industrial communication systems (ICS) are undergoing a transformation, leveraging a multitude of technologies to meet the specific needs of the manufacturing and automation industries. The convergence of these networks into edge, fog, and cloud architectures has enhanced their scalability and facilitated the deployment of advanced data-driven approaches, such as machine learning for optimizing production processes. However, ensuring proper provisioning of network and computation resources, along with delivering quality of service, is increasingly challenging in these complex communication systems. Zero-Touch Management (ZTM) frameworks promise to reduce complexity and minimize dependence on manual configuration by human experts. Successful deployment of such frameworks requires an accurate Network Digital Twin (NDT) of relevant network elements, as autonomous decision-making heavily relies on the quantity and quality of historical and real-time node and link state information provided by the NDT. However, the use of NDTs for ICS and ZTM in particular is still an emerging research topic. This paper therefore proposes a theoretical use-case for an NDT-based ZTM framework to improve resource utilization in cloud-centered networks. It presents a state-of-the-art analysis of recent NDT advances enabling the deployment of related ZTM approaches and discusses associated challenges and future research directions.

Increasing traffic congestion leads to significant costs, whereby poorly configured signaled intersections are a common bottleneck and root cause. Traditional traffic signal control (TSC) systems employ rule-based or heuristic methods to decide signal timings, while adaptive TSC solutions utilize a traffic-actuated control logic to increase their adaptability to real-time traffic changes. However, such systems are expensive to deploy and are often not flexible enough to adequately adapt to the volatility of today’s traffic dynamics. More recently, this problem became a frontier topic in the domain of deep reinforcement learning (DRL) and enabled the development of multi-agent DRL approaches that can operate in environments with several agents present, such as traffic systems with multiple signaled intersections. However, many of these proposed approaches were validated using artificial traffic grids. This paper presents a case study, where real-world traffic data from the town of Lemgo in Germany is used to create a realistic road model within VISSIM. A multi-agent DRL setup, comprising multiple independent deep Q-networks, is applied to the simulated traffic network. Traditional rule-based signal controls, modeled in LISA+ and currently employed in the real world at the studied intersections, are integrated into the traffic model and serve as a performance baseline. The performance evaluation indicates a significant reduction of traffic congestion when using the RL-based signal control policy over the conventional TSC approach with LISA+. Consequently, this paper reinforces the applicability of RL concepts in the domain of TSC engineering by employing a highly realistic traffic model.

Over the past decades industrial communication networks have evolved into highly diverse and heterogeneous environments, with a variety of different technologies being deployed to address the diverse requirements of manufacturing-and automation-specific use cases. These include stringent latency limits, high availability and reliability, as well as deterministic communication behavior. To assure the necessary allocation of re-sources and provisioning of required Quality-of-Service in highly diverse communication systems, a holistic network management approach is needed that can serve all cornerstones of modern industrial networks. More recently, this lead to the development of new adaptive and agile management approaches that imple-ment autonomous and self-organizing manufacturing networks, whereby Machine Learning (ML) methods started to become an integral part for overcoming the limiting factors of practically deploying such systems. Due to the growing complexity of today's networking environments, defining network management policies based on expert knowledge becomes increasingly difficult. ML has evolved as a promising technique to extract knowledge from collected data to enable cognitive network management approaches. This paper reviews past advances in ML applications for zero touch management of heterogeneous industrial communication networks. It illustrates how a network's management life-cycle that is based on digital twin technology can harnesses the potentials of ML to bring the concepts of organic computing and zero-touch cognitive manufacturing within industrial networks closer to reality. Lastly, recent papers that discuss the use of ML approaches for self-x features in Zero-Touch Management (ZTM) network environments are surveyed and relevant open issues are discussed.

Sub-optimal control policies in intersection traffic signal controllers (TSC) contribute to congestion and lead to negative effects on human health and the environment. Reinforcement learning (RL) for traffic signal control is a promising approach to design better control policies and has attracted considerable research interest in recent years. However, most work done in this area used simplified simulation environments of traffic scenarios to train RL-based TSC. To deploy RL in real-world traffic systems, the gap between simplified simulation environments and real-world applications has to be closed. Therefore, we propose LemgoRL, a benchmark tool to train RL agents as TSC in a realistic simulation environment of Lemgo, a medium-sized town in Germany. In addition to the realistic simulation model, LemgoRL encompasses a traffic signal logic unit that ensures compliance with all regulatory and safety requirements. LemgoRL offers the same interface as the well-known OpenAI gym toolkit to enable easy deployment in existing research work. To demonstrate the functionality and applicability of LemgoRL, we train a state-of-the-art Deep RL algorithm on a CPU cluster utilizing a framework for distributed and parallel RL and compare its performance with other methods. Our benchmark tool drives the development of RL algorithms towards real-world applications.

The Plug-and-Produce concept requires that after connecting a new module to a system, the exchange of the configuration data takes place. As further operation of the system depends on this initialization procedure, it is necessary to ensure that the data presented by the system and the newly attached component is authentic. Therefore, we propose a new concept for secure Plug-and-Produce functionality, which exploits the combination of the Asset Administration Shell (AAS) and Blockchain technology. On the one hand, the AAS shall be responsible for presenting uniform and standardized configuration data as well as for storing and managing Blockchain. On the other, Blockchain shall ensure authenticity and integrity of the configuration data.