Maryam Vahabi

Modelling plays a critical role in many engineering applications. Partial differential equations (PDEs) are ubiquitous, describing various physical phenomena such as fluid flow, electromagnetism, and quantum mechanics. Although some of these equations have analytical solutions, many require high-fidelity simulations of parametric PDEs. In general, high-fidelity simulations are computationally expensive and often infeasible for real-time or multi-query applications. This challenge has led to the development of reduced-order models (ROMs). Over the past few decades, ROMs have emerged as a practical solution for simulating, controlling, and optimizing large-scale and complex dynamical systems. This paper introduces a novel Calibrated Intrusive Reduced-Order Modelling (CIROM) approach for the efficient and accurate simulation of the one-dimensional Burgers’ equation, employed as a canonical benchmark because it is a simplified fundamental partial differential equation that captures the behaviour of many real-world phenomena. The proposed method, combining the strengths of proper orthogonal decomposition (POD) and long short-term memory (LSTM) networks, effectively reduces computational complexity while addressing inherent instabilities in classical reduced-order models. Unlike traditional POD-ROMs, which often suffer from error accumulation and instability at high Reynolds numbers, the CIROM employs an iterative LSTM-based error correction mechanism to learn and compensate for truncation and projection errors. This study is benchmark-oriented and does not aim to provide a general PDE solver. The performance of the proposed method is rigorously evaluated across a broad range of Reynolds numbers, including interpolation and extrapolation scenarios, demonstrating robust extrapolation within moderate ranges. Comprehensive numerical experiments confirm that the CIROM outperforms both pure intrusive ROMs and purely data-driven LSTM models in terms of prediction accuracy, stability, and computational cost.

In cyber-physical systems governed by nonlinear partial differential equations (PDEs), real-time control is often limited by sparse sensor data and high-dimensional system dynamics. Deep reinforcement learning (DRL) has shown promise for controlling such systems, but training DRL agents directly on full-order simulations is computationally intensive. This paper presents a sensor-driven, non-intrusive reduced-order modeling (NIROM) framework called FAE-CAE-LSTM, which combines convolutional and fully connected autoencoders with a long short-term memory (LSTM) network. The model compresses high-dimensional states into a latent space and captures their temporal evolution. A DRL agent is trained entirely in this reduced space, interacting with the surrogate built from sensor-like spatiotemporal measurements, such as pressure and velocity fields. A CNN-MLP reward estimator provides data-driven feedback without requiring access to governing equations. The method is tested on benchmark systems including Burgers’ equation, the Kuramoto–Sivashinsky equation, and flow past a circular cylinder; accuracy is further validated on flow past a square cylinder. Experimental results show that the proposed approach achieves accurate reconstruction, robust control, and significant computational speedup over traditional simulation-based training. These findings confirm the effectiveness of the FAE-CAE-LSTM surrogate in enabling real-time, sensor-informed, scalable DRL-based control of nonlinear dynamical systems.

Industrial Internet of Things (IIoT) is changing the modern industrial processes through the integration of devices, sensors, and control systems, enabling real-time monitoring, automation, and datadriven decision-making. A significant challenge within IIoT systems, especially in control applications, is achieving guaranteed latency while ensuring high reliability. This paper presents a comprehensive review of recent research in IIoT, with a particular emphasis on the role of fifth generation (5G) wireless communication technologies. It includes an in-depth analysis of experiments conducted via real-world test-beds, providing insights into the performance of IIoT systems utilizing 5G technology. Additionally, the important support which has taken from different simulations and emulations in terms of library, protocols stacks and so on also discussed. We also assess how effectively these test-beds and platforms replicate real-world industrial environments and their capability to evaluate system performance under varying network configurations. Furthermore, we explore key technical parameters commonly used in IIoT research, including throughput, jitter, packet loss, spectral efficiency, and energy efficiency. This paper highlights both the benefits and limitations of employing test-beds for evaluating IIoT systems in industrial applications. It further examines the essential role of futuristic technologies, challenges and outlines research directions aimed at IIoT systems.

RPL—Routing Protocol for Low-Power and Lossy Networks (usually pronounced “ripple”)—is the de facto standard for IoT networks. However, it neglects to exploit IoT devices’ full capacity to optimize their transmission power, mainly because it is quite challenging to do so in parallel with the routing strategy, given the dynamic nature of wireless links and the typically constrained resources of IoT devices. Adapting the transmission power requires dynamically assessing many parameters, such as the probability of packet collisions, energy consumption, the number of hops, and interference. This paper introduces Adaptive Control of Transmission Power for RPL (ACTOR) for the dynamic optimization of transmission power. ACTOR aims to improve throughput in dense networks by passively exploring different transmission power levels. The classic solutions of bandit theory, including the Upper Confidence Bound (UCB) and Discounted UCB, accelerate the convergence of the exploration and guarantee its optimality. ACTOR is also enhanced via mechanisms to blacklist undesirable transmission power levels and stabilize the topology of parent–child negotiations. The results of the experiments conducted on our 40-node, 12-node testbed demonstrate that ACTOR achieves a higher packet delivery ratio by almost 20%, reduces the transmission power of nodes by up to 10 dBm, and maintains a stable topology with significantly fewer parent switches compared to the standard RPL and the selected benchmarks. These findings are consistent with simulations conducted across 7 different scenarios, where improvements in end-to-end delay, packet delivery, and energy consumption were observed by up to 50%.

Internet-of-Things (IoT) applications are envisaged to evolve to support mobility of devices while providing quality of service in the system. To keep the connectivity of the constrained nodes upon topological changes, it is of vital importance to enhance the standard protocol stack, including the Routing Protocol for Lossy Low-power Networks (RPL), with accurate and real-time control decisions. We argue that devising a centralized mobility management solution based on a lightweight Software Defined Networking (SDN) controller provides seamless handoff with reasonable communication overhead. A centralized controller can exploit its global view of the network, computation capacity, and flexibility, to predict and significantly improve the responsiveness of the network. This approach requires the controller to be fed with the required input and to get involved in the distributed operation of the standard RPL. We present SDMob, which is a lightweight SDN-based mobility management architecture that integrates an external controller within a constrained IoT network. SDMob lifts the burden of computation-intensive filtering algorithms away from the resource-constrained nodes to achieve seamless handoffs upon nodes’ mobility. The current work extends our previous work, by supporting multiple mobile nodes, networks with a high density of anchors, and varying hop-distance from the controller, as well as harsh and realistic mobility patterns. Through analytical modeling and simulations, we show that SDMob outperforms the baseline RPL and the state-of-the-art ARMOR in terms of packet delivery ratio and end-to-end delay, with an adjustable and tolerable overhead. With SDMob, the network provides close to 100% packet delivery ratio (PDR) for a limited number of mobile nodes, and maintains sub-meter accuracy in localization under random mobility patterns and varying network topologies.

There are many initiatives and technologies working towards implementing factories of the future. One consensus is that the classical hierarchical automation system design needs to be flattened while supporting the functionality of both Operation Technology (OT) and Information Technology (IT) within the same network infrastructure. To achieve the goal of IT/OT convergence in process automation, an evolutionary transition is preferred. Challenges are foreseen during the transition, mainly caused by the traditional automation architecture, and the main challenge is to identify the gap between the current and future network architectures. To address the challenges, in this paper, we describe one desired future scenario for process automation and carry out traffic measurements from a pulp and paper mill. The measured traffic is further analyzed, which reveals representative traffic characteristics in the process automation. Finally, the key challenges and future directions towards a system architecture for factories of the future are presented.

Recent advances in artificial intelligence and machine learning (ML) led to effective methods and tools for analyzing the human behavior. Human Activity Recognition (HAR) is one of the fields that has seen an explosive research interest among the ML community due to its wide range of applications. HAR is one of the most helpful technology tools to support the elderly’s daily life and to help people suffering from cognitive disorders, Parkinson’s disease, dementia, etc. It is also very useful in areas such as transportation, robotics and sports. Deep learning (DL) is a branch of ML based on complex Artificial Neural Networks (ANNs) that has demonstrated a high level of accuracy and performance in HAR. Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are two types of DL models widely used in the recent years to address the HAR problem. The purpose of this paper is to investigate the effectiveness of their integration in recognizing daily activities, e.g., walking. We analyze four hybrid models that integrate CNNs with four powerful RNNs, i.e., LSTMs, BiLSTMs, GRUs and BiGRUs. The outcomes of our experiments on the PAMAP2 dataset indicate that our proposed hybrid models achieve an outstanding level of performance with respect to several indicative measures, e.g., F-score, accuracy, sensitivity, and specificity.