The maritime industry has a significant influence on the global economy, underscoring the need for operational availability and safety through effective maintenance practices. Predictive maintenance emerges as a promising solution compared to conventional maintenance schemes currently employed by the industry, offering proactive failure predictions, reduced downtime events, and extended machinery lifespan. This paper addresses a critical gap in the existing literature by providing a comprehensive overview of the main data-driven PdM systems. Specifically, the review explores common issues found in vessel components (i.e., propulsion, auxiliary, electric, hull), examining how different state-of-the-art PdM architectures, ranging from basic machine learning models to advanced deep learning techniques aim to address them. Additionally, the concepts of centralized machine learning, federated, and transfer learning are also discussed, demonstrating their potential to enhance PdM systems as well as their limitations. Finally, the current challenges hindering adoption are discussed, together with the future directions to advance implementation in the field.

Current and upcoming data-intensive Mission Critical (MC) applications rely on high Quality of Service (QoS) requirements related to connectivity, latency and network reliability. Beyond 5G networks shall accommodate MC services that enable voice, data and video transfer in extreme circumstances, for instance in occurrence of network overloads or infrastructure failures. In this work, we describe the specifications of the architectural framework that enables the roll-out of MC services over 5G networks and beyond, considering recent technological advancements of cloud-native functionalities, network slicing and edge deployments. The network architecture and the deployment process is described in three practical scenarios, including a capacity increase in the service load that necessitates the scaling of the computational resources, the deployment of a dedicated network slice for accommodating the stringent requirement of a MC application and a service migration scenario at the edge to cope with critical failures and QoS degradation. Furthermore, we illustrate the implementation of a Machine Learning (ML) algorithm that is used for overload prediction, validating its ability to predict the capacity increase and notify the components responsible to trigger the appropriate actions, based on a real dataset. To this end, we mathematically define the overload detection problem, as well as generalized prediction tasks in emergency situations and examine the key parameters (proactiveness ability, loockback window, etc.) of the ML model, also comparing its predictions abilities (similar to 93% accuracy in overload detection) against multiple baseline classifiers. Finally, we demonstrate the flexibility of the ML model to achieve reliable predictions in scenarios with diverse requirements.

The evolution towards sixth-generation (6G) networks requires new architecture enhancements to support the broad device ecosystem, comprising users, machines, autonomous vehicles, and Internet-of-things devices. Moreover, high heterogeneity in the desired quality-of-service (QoS) is expected, as 6G networks will offer extremely low-latency and high-throughput services and error-free communication. This complex environment raises significant challenges in resource management while adhering to security and privacy constraints due to the plethora of data generation endpoints. Considering the advances in AI/ML-aided integration in wireless networks and recent efforts on the network data analytics function (NWDAF) by the 3rd generation partnership project (3GPP), this work presents an AI/ML-aided distributed trustable engine (DTE), collecting data from diverse sources of the 6G infrastructure and deploying ML methods for anomaly detection against diverse threat types. Moreover, we present the DTE architecture and its components, providing data management, AI/ML model training, and classification capabilities for anomaly detection. To promote privacy-aware networking, a federated learning (FL) framework to extend the DTE is discussed. Then, the anomaly detection capabilities of the AI/ML-aided DTE are presented in detail, together with the ML model training process, which considers various ML models. For this purpose, we use two open datasets representing attack scenarios in the core and the edge parts of the network. Experimental results, including an ensemble learning method and different supervised learning alternatives, show that the AI/ML-aided DTE can efficiently train ML models with reduced dimensionality and deploy them in diverse cybersecurity scenarios to improve anomaly detection in 6G networks.

The energy consumed by buildings is expected to significantly rise in the upcoming years, necessitating intelligent Home Energy Management Systems (HEMS) that create comfortable conditions for their inhabitants, while also offering sustainable and cost-effective solutions. The building environment, however, includes multiple time-varying parameters that cannot be controlled, such as the output of renewable energy sources, the market-dependent electricity prices, the outdoor temperature, as well as the occupants' energy habits. To overcome these barriers, we propose a hybrid Machine Learning (ML) algorithm for smart HEMS control, leveraging the properties of a decision-making deep deterministic policy gradient model, enhanced by the predictive capabilities of long short-term memory networks. Hence, the proposed algorithm aims to achieve an optimal balance between energy cost and occupant comfort by continuously adjusting the energy provided to the heating, ventilation, and air conditioning system, as well as controlling the energy storage system of the smart home. The proposed hybrid method is validated with simulations using real-world data and compared against baseline approaches, showcasing its effectiveness to achieve an optimal trade-off between the indoor temperature deviation and the average energy cost. Note to Practitioners-This paper is motivated by the need of automated energy management in smart buildings to achieve minimal energy consumption. Continuous demands for heating, ventilation and air conditioning, as well as utilization of household loads lead to a significant increase in the daily energy consumption. This paper explores the trade-off between the price that a household pays in return for indoor temperature comfort and coverage of the energy requirements. For this purpose, we provide a comprehensive mathematical modeling of a smart home environment, followed by the description of a real-time decision-making energy management algorithm that is based on Machine Learning (ML). This algorithm takes into account the forecasting of energy consumption and energy production of the household in the future, optimally adjusting the power that is drawn from the grid. By implementing this algorithm using real-world data, we demonstrate that a smart home can maintain a comfortable indoor temperature, while minimizing the energy cost and boosting the use of local renewable energy. In this context, manufacturers, developers of smart home systems or energy providers that specialize in building automation systems can utilize this algorithm to offer consumers automated energy management that optimizes cost and comfort, leading to sustainable operations. We plan to extend this model to energy communities consisting of multiple smart homes that can also exchange energy surplus to achieve zero-sum energy footprint.

The acoustic startle reflex (ASR) relies on the sensorimotor system and is affected by aging, sex, and psychopathology. ASR can be modulated by the prepulse inhibition (PPI) paradigm, which achieves the inhibition of reactivity to a startling stimulus (pulse) following a weak prepulse stimulus. Additionally, neurophysiological studies have found that brain activity is characterized by irregular patterns with high complexity, which however reduces with age. Our study investigated the relationship between pre-startle nonlinear dynamics and prepulse inhibition in healthy children vs. adults. Fifty-six individuals took part in the experiment: 31 children and adolescents and 25 adults. Participants heard 51 pairs of tones (prepulse, startle) with a time difference of 30 to 500 ms. Subsequently, we assessed neural complexity by computing the largest Lyapunov exponent (LLE) during the pre-startle period and assessed PPI by analyzing the post-startle event-related potentials (ERPs). Results showed higher neural complexity for children compared to adults, in line with previous research showing reduced complexity in the physiological signals in ageing. As expected, PPI (as reflected in the P50 and P200 components) was enhanced in adults compared to children, potentially due to the maturation of the ASR for the former. Interestingly, pre-startle complexity was correlated with the P50 component in children only, but not in adults, potentially due to the different stage of sensorimotor maturation between age groups. Overall, our study offers novel contributions for investigating brain dynamics, linking nonlinear with linear measures. Our findings are consistent with the loss of neural complexity in ageing, and suggest differentiated links between nonlinear and linear metrics in children and adults.

Open Radio Access Network (O-RAN) is transforming the telecommunications landscape by enabling flexible, intelligent, and multi-vendor networks. Central to its architecture are xApps hosted on the Near-Real-Time RAN Intelligent Controller (Near-RT RIC), which optimize network functions in real time. However, the concurrent operation of multiple xApps with conflicting objectives can lead to suboptimal performance. This paper introduces a generalized Conflict Management scheme for Multi-Channel Power Control in O-RAN xApps (COMIX), designed to detect and resolve conflicts between xApps. To demonstrate COMIX, we focus on two Deep Reinforcement Learning (DRL)-based xApps for power control: one maximizes the data rare across UEs, and the other optimizes system-level energy efficiency. COMIX employs a standardized Conflict Mitigation Framework (CMF) for conflict detection and resolution and leverages the Network Digital Twin (NDT) to evaluate the impact of conflicting actions before applying them to the live network. We validate the framework using a realistic multi-channel power control scenario under various conflict resolution policies, demonstrating its effectiveness in balancing antagonistic objectives. Evaluation results show that COMIX achieves up to 60% energy savings across different Service-Level Agreement (SLA) policies compared to a baseline conflict-unaware system, with negligible impact (around 3%) on system throughput. While this study considers power control xApps, the COMIX framework is generalizable and can be applied to any xApp conflict scenario involving resource contention or KPI interdependence.

Maritime transportation is crucial for global trade and responsible for the majority of goods movement worldwide. The optimization of maritime operations is challenged by the complexity and heterogeneity of maritime nodes. This paper presents the emerging deployment of federated learning (FL) in maritime environments to address these challenges. FL enables decentralized machine learning model training, ensuring data privacy and security while overcoming issues associated with non-i.i.d. data. This paper explores various maritime use cases, including fuel consumption reduction, predictive maintenance, and just-in-time arrival. Experimental results using real datasets demonstrate the superiority of FL in predicting the fuel consumption of large cargo ships in terms of accuracy and spatiotemporal complexity over traditional collaborative machine learning approaches. The findings indicate that FL can significantly improve the performance of fuel consumption models in a collaborative way, while ensuring data privacy preservation and no data transmission during the learning process. Finally, this paper discusses open issues and future research directions necessary for the widespread adoption of FL in maritime transportation and settings.

Maritime and shipping are unambiguously the cornerstones of the global economy and transportation. To improve efficiency, maritime sector activities are focused on the realization of Smart Shipping (SMS), leveraging 6G Communications, Energy Efficiency (EE) and Machine Learning (ML). However, conventional Centralized Machine Learning (CML) cannot be easily applied in the maritime, mainly due to the drawbacks: (i) prohibitive data communication overhead and bandwidth limitations, since CML requires centralization of massive data through transmissions from heterogeneous sources, (ii) excessive energy consumption associated with massive data transfers, (iii) remarkable transmission errors due to harsh propagation conditions, and (iv) data privacy violation, since the data carries sensitive and commercial information. This article proposes a two-fold Federated Learning (FL) scheme (FedShip) to improve the privacy, EE and communication-efficiency of future 6G maritime networks. FedShip uses the Over-the-Air computation (AirComp) principles to exploit the signal superposition property and ensure that local models are accurately and efficiently combined. Using real data regarding the fuel consumption of multiple cargo ships, we compared the FL performance, building multiple timeseries forecasting models, with collaborative ML baselines. AirComp performance was also assessed using simulation data about channel measurements. After optimizing the hyperparameters of the local models, extensive results revealed that: (i) FL shows enhanced fuel prediction accuracy (95.5% relative to the CML), while ensuring data privacy and (ii) AirComp can be adopted to combine the local models with low computation error, offering significant EE and spectrum efficiency improvements, especially when dense 6G scenarios are considered.

The next generation of shipping industry, namely Shipping 4.0 will integrate advanced automation and digitization technologies towards revolutionizing the maritime industry. As conventional maintenance practices are often inefficient, costly, and unable to cope with unexpected failures, leading to operational disruptions and safety risks, the need for efficient predictive maintenance (PdM), relying on machine learning (ML) algorithms is of paramount importance. Still, the exchange of training data might raise privacy concerns of the involved stakeholders. Towards this end, federated learning (FL), a decentralized ML approach, enables collaborative model training across multiple distributed edge devices, such as on-board sensors and unmanned vessels and vehicles. In this work, we explore the integration of FL into PdM to support Shipping 4.0 applications, by using real datasets from the maritime sector. More specifically, we present the main FL principles, the proposed workflow and then, we evaluate and compare various FL algorithms in three maritime use cases, i.e. regression to predict the naval propulsion gas turbine (GT) measures, classification to predict the ship engine condition, and time-series regression to predict ship fuel consumption. The efficiency of the proposed FL-based PdM highlights its ability to improve maintenance decision-making, reduce downtime in the shipping industry, and enhance the operational efficiency of shipping fleets. The findings of this study support the advancement of PdM methodologies in Shipping 4.0, providing valuable insights for maritime stakeholders to adopt FL, as a viable and privacy-preserving solution, facilitating model sharing in the shipping industry and fostering collaboration opportunities among them.

This paper addresses the application of neural networks in resource constrained edge-devices. The goal is to achieve a speedup both in inference and training time, with minimal accuracy loss. More specifically, it brings to light the need for compressing current models, which are mostly developed with access to more resources that the device that the model will potential run on. With the recent advances of Internet of Things(IoT) the number of devices has and is expected to rise. Not only are these devices computationally limited, but their capabilities are nor homogeneous nor predictable at the time of the development of a model, as new devices can be added anytime. This creates the need to quickly and efficiently produce models that fit each devices specifications. Transfer learning is a very efficient method, in terms of training time, but confines the user to the dimensionality of the pretrained model. Pruning is used as a way to overcome this obstacle and carry over knowledge to a variety of model, that differ in size. The aim of this paper is to serve as an introduction to pruning as a concept, as a template for further research, quantify the efficiency of a variety of methods and expose some of it's limitations. Pruning was performed on a telecommunications anomaly dataset and the results were compared to a baseline, in regards to speed and accuracy.

Alpha oscillations are known to play a central role in several higher-order cognitive functions, especially selective attention, working memory, semantic memory, and creative thinking. Nonetheless, we still know very little about the role of alpha in the generation of more remote semantic associations, which is key to creative and semantic cognition. Furthermore, it remains unclear how these oscillations are shaped by the intention to “be creative,” which is the case in most creativity tasks. We aimed to address these gaps in two experiments. In Experiment 1, we compared alpha oscillatory activity (using a method which distinguishes genuine oscillatory activity from transient events) during the generation of free associations which were more vs. less distant from a given concept. In Experiment 2, we replicated these findings and also compared alpha oscillatory activity when people were generating free associations versus associations with the instruction to be creative (i.e. goal-directed). We found that alpha was consistently higher during the generation of more distant semantic associations, in both experiments. This effect was widespread, involving areas in both left and right hemispheres. Importantly, the instruction to be creative seems to increase alpha phase synchronisation from left to right temporal brain areas, suggesting that intention to be creative changed the flux of information in the brain, likely reflecting an increase in top-down control of semantic search processes. We conclude that goal-directed generation of remote associations relies on top-down mechanisms compared to when associations are freely generated.

The deployment of maritime communication networks (MCNs) enables Internet-of-Things (IoT) applications, related to autonomous navigation, offshore facilities and smart ports. Still, the majority of maritime nodes, residing in MCNs lacks reliable connectivity. Towards this end, integrating unmanned aerial vehicles (UAVs) in sixth generation (6G) MCN topologies results in the formation of an aerial segment, complementing shore base stations that may offer insufficient coverage, and satellite communication, characterized by increased delays. In this study, we focus on an MCN where the direct links towards a shore BS are not available, due to excessive fading conditions. For this case, we use a UAV swarm to provide improved wireless connectivity, adopting non-orthogonal multiple access (NOMA) for high resource efficiency. In downlink communication, UAVs take into consideration the desired service rate and the channel quality of their links towards the maritime nodes. In the uplink, UAVs employ dynamic decoding ordering to enhance the performance of successive interference cancellation, avoiding fixed ordering of the maritime nodes' signals. Moreover, to ensure highly flexible UAV selection, UAVs are equipped with buffers to store data. Performance comparisons show that the UAV swarm-aided MCN enjoys increased average sum-rate by relying on multi-criteria-based interference cancellation and buffer-aided UAVs, over other benchmark schemes in the downlink and uplink. Finally, the delay-aware nature of the proposed algorithms where the UAV-destination links are prioritized, leads to reduced average delay.

In the burgeoning domain of the edge-cloud con-tinuum (ECC), the efficient management of computational tasks offloaded from mobile devices to edge nodes is paramount. This paper introduces a Cooperative cOmputation Offloading scheme for ECC via Latency-aware multi-agent Reinforcement learning (COOLER), a distributed framework designed to address the challenges posed by the uncertain load dynamics at edge nodes. COOLER enables each edge node to autonomously make offloading decisions, optimizing for non-divisible, delay-sensitive tasks without prior knowledge of other nodes‘ task models and decisions. By formulating a multi-agent computation offloading problem, COOLER aims to minimize the expected long-term latency and task drop ratio. Following the ECC requirements for seamless task flow both within Edge layer and between Edge-Cloud layers, COOLER considers that task computation decisions are three-fold: (i) local computation, (ii) horizontal offloading to another edge node, or (iii) vertical offloading to the Cloud. The integration of advanced techniques such as long short-term memory (LSTM), double deep Q-network (DQN) and dueling DQN enhances the estimation of long-term costs, thereby improving decision-making efficacy. Simulation results demonstrate that COOLER significantly outperforms baseline offloading algorithms, reducing both the ratio of dropped tasks and average delay, and better harnessing the processing capacities of edge nodes.

The maritime domain is a major driver of economic growth with emerging services, comprising intelligent transportation systems (ITSs), smart ports, security and safety, and ocean monitoring systems. Sixth generation (6G) mobile networks will offer various technologies, paving the way for reliable and autonomous maritime communication networks (MCNs), supporting these novel maritime services. This review presents the main enabling technologies for future MCNs and relevant use cases, including ITSs with reduced carbon footprint, ports and maritime infrastructure security, as well as fault detection and predictive maintenance. Moreover, the current trends in integrated satellite-aerial-terrestrial-maritime network architectures are discussed together with the different network segments and communication technologies, and machine learning integration aspects.

Introduction: Shipping and maritime transportation have gradually gained a key role in worldwide economical strategies and modern business models. The realization of Smart Shipping (SMS) powered by advanced 6G communication networks, as well as innovative Machine Learning (ML) solutions, has recently become the focal point in the maritime sector. However, conventional centralized learning schemes are unsuitable in the maritime domain, due to considerable data communication overhead, stringent energy constraints, increased transmission failures in the harsh propagation environment, as well as data privacy concerns.Methods: To overcome these challenges, we propose the joint adoption of Federated Learning (FL) principles and the utilization of the Over-the-Air computation (AirComp) wireless transmission framework. Thus, this paper initially describes the mathematical considerations of a 6G maritime communication system, focusing on the heterogeneity of the relevant nodes and the channel models, including an Unmanned Aerial Vehicle (UAV)-aided relaying model that is usually required in maritime communications. The communication network, enhanced with the AirComp technique for efficiency purposes, forms the technical basis for the collaborative learning across multiple Internet of Maritime Things (IoMT) nodes in FL tasks. The workflow of the FL/AirComp scheme is illustrated and proposed as a communication-efficient and privacy-aware SMS framework, considering spectrum and energy efficiency aspects under a sum transmitting power constraint.Results: Then, the performance of the proposed methodology is assessed in an important ML task, related to intelligent maritime transportation systems, namely, the prediction of the Cargo Ship Propulsion Power using real data originating from six cargo ships and utilizing long-short-term-memory (LSTM) neural networks. Upon extensive experimentation, FL showed higher prediction accuracy relative to the typical Ensemble Learning technique by a factor of 3.04. The AirComp system performance was evaluated under varying noise conditions and number of IoMT nodes, using simulation data for the channel state information by regulating the power of the transmitting IoMT entities and the scaling factor at the shore base station.Discussion: The results clearly indicate the efficiency of the proposed FL/AirComp scheme in achieving low computation error, collaborative learning, spectrum efficiency and privacy protection in wireless maritime communications, while providing adequate accuracy levels with respect to the optimization objective.

In this paper, a novel IoT-Edge-Cloud (IEC) computing system designed for multiple Smart Homes is introduced, with a focus on supporting Home Energy Management Systems (HEMS) for temperature control within a defined comfort range. Leveraging model-free deep reinforcement learning, the proposed method, Smart Home Energy and Temperature Control (SHETEC), employs autonomous agents which are trained to manipulate the input power of Heating, Ventilation, and Air Conditioning (HVAC) systems and charging/discharging power of Energy Storage Systems (ESS) using Deep Deterministic Policy Gradients (DDPG). In addition, we present the Average Opinion (AO) method, a collaborative decision-making approach that combines the models of all Smart Homes in a distributed approach. Experimental results, conducted through simulation on three Smart Homes using real-world heterogeneous data, demonstrate the effectiveness of both SHETEC and Average Opinion in maintaining temperatures within the desired comfort bounds.

As World Economic Forum has stated about Artificial Intelligence 1 , 2 , characterised often as the Fourth Industrial Revolution, "the speed, breadth and depth of this revolution is forcing society to rethink how countries develop, how organisations create value and even what it means to be human". However, justice and law enforcement are sectors that don't attract the attention and the focus of innovation, as much as other sectors do, not unreasonably, concerning the complexity of the field, the legal implications and the humancentred character of the legal science and the legal professions. Nevertheless, the recent developments of AI systems have opened new avenues to support and modernise the judicial system to enhance its efficiency and to better serve the grant of justice. In the present research we explore the case of automated decision-making solutions, used to support the judge's task or even to replace him/her, focusing on the legal requirements for their implementation, the challenges that they pose and the adequate mitigation measures to address these challenges. Our research concerns the European civil jurisdiction, since in the field of criminal justice, algorithmic driven decisions pose different challenges that need distinct consideration, concerning mostly the specific rights of the defendant. As far as administrative justice is concerned, the remarks of our research may have equivalent application.

Cloud-Edge Computing Continuum (CEC) system, where edge and cloud nodes are seamlessly connected, is dedicated to handle substantial computational loads offloaded by end-users. These tasks can suffer from delays or be dropped entirely when deadlines are missed, particularly under fluctuating network conditions and resource limitations. The CEC is coupled with the need for hybrid task offloading, where the task placement decisions concern whether the tasks are processed locally, offloaded vertically to the cloud, or horizontally to interconnected edge servers. In this paper, we present a distributed hybrid task offloading scheme (HOODIE) designed to jointly optimize the tasks latency and drop rate, under dynamic CEC traffic. HOODIE employs a model-free deep reinforcement learning (DRL) framework, where distributed DRL agents at each edge server autonomously determine offloading decisions without global task distribution awareness. To further enhance the system pro-activity and learning stability, we incorporate techniques such as Long Short-term Memory (LSTM), Dueling deep Q-networks (DQN), and double-DQN. Extensive simulation results demonstrate that HOODIE effectively reduces task drop rates and average task processing delays, outperforming several baseline methods under changing CEC settings and dynamic conditions.

Maritime trade unavoidably influences the economy, transportation and market worldwide. Deploying efficient, privacy-preserving and environmental-friendly solutions in maritime environments requires a continuous global effort. There are many obstacles associated with the centralized optimization for on-sail ships, including data security violation, signaling overhead and high-latency response. Contradictorily, the emerging Federated Learning (FL) paradigm has been proposed as an efficient solution for promoting data privacy, low latency and high communication efficiency. In this context, this paper proposes a maritime-based FL scheme to ensure distributed, collaborative and secure optimization for enabling predictive on-ship intelligence. A scenario targeting the accurate prediction of the Primary Engine Power (PEP) of large cargo ships is considered, as the output PEP is directly proportional to fuel consumption. Based on real data, several Machine Learning (ML) models were tested and validated, in terms of their PEP prediction accuracy, with all models exploiting both weather- and ship-related data. Simulation results towards the selection of the PEP predictor, as well as the quantification of the FL scheme efficacy over traditional benchmarks were carried out, achieving a beneficial equilibrium between the prediction accuracy, data privacy and communication efficiency.