Explore the vast range of titles available.

Given the increasing energy demand and the environmental consequences of fossil fuel consumption, the shift toward sustainable energy sources has become a global priority. Renewable Energy Communities (RECs)—comprising citizens, businesses, and legal entities—are emerging to democratise access to renewable energy. These communities allow members to produce their own energy, sharing or selling any surplus, thus promoting sustainability and generating economic value. However, scaling RECs while ensuring profitability is challenging due to renewable energy intermittency, price volatility, and heterogeneous consumption patterns. To address these issues, this paper presents a Machine Learning as a Service (MLaaS) framework, where each REC microgrid has a customised Reinforcement Learning (RL) agent and electricity price forecasts are included to support decision-making. All the conducted experiments, using the open-source simulator Pymgrid, demonstrate that the proposed agents reduced operational costs by up to 96.41% compared to a robust baseline heuristic. Moreover, this study also introduces two cost-saving features: Peer-to-Peer (P2P) energy trading between communities and internal energy pools, allowing microgrids to draw local energy before using the main grid. Combined with the best-performing agents, these features achieved trading cost reductions of up to 45.58%. Finally, in terms of deployment, the system relies on an MLOps-compliant infrastructure that enables parallel training pipelines and an autoscalable inference service. Overall, this work provides significant contributions to energy management, fostering the development of more sustainable, efficient, and cost-effective solutions.

The increasing demand for distributed machine learning like Federated Learning (FL) in dynamic, resource-constrained edge environments, 5G/6G networks, and the proliferation of mobile and edge devices, presents significant challenges related to fault tolerance, elasticity, and communication efficiency. This paper addresses these issues by proposing a novel modular and resilient FL framework. In this context, resilience refers to the system’s ability to maintain operation and performance despite disruptions. The framework is built on decoupled modules handling core FL functionalities, allowing flexibility in integrating various algorithms, communication protocols, and resilience strategies. Results demonstrate the framework’s ability to integrate different communication protocols and FL paradigms, showing that protocol choice significantly impacts performance, particularly in high-volume communication scenarios, with Zenoh and MQTT exhibiting lower overhead than Kafka in tested configurations, and Zenoh emerging as the most efficient communication option. Additionally, the framework successfully maintained model training and achieved convergence even when simulating probabilistic worker failures, achieving a MCC of 0.9453.

The Internet of Vehicles (IoV) presents complex cybersecurity challenges, particularly against Denial-of-Service (DoS) and spoofing attacks targeting the Controller Area Network (CAN) bus. This study leverages the CICIoV2024 dataset, comprising six distinct classes of benign traffic and various types of attacks, to evaluate advanced machine learning techniques for instrusion detection systems (IDS). The models XGBoost, Random Forest, AdaBoost, Extra Trees, Logistic Regression, and Deep Neural Network were tested under realistic, imbalanced data conditions, ensuring that the evaluation reflects real-world scenarios where benign traffic dominates. Using hyperparameter optimization with Optuna, we achieved significant improvements in detection accuracy and robustness. Ensemble methods such as XGBoost and Random Forest consistently demonstrated superior performance, achieving perfect accuracy and macro-average F1-scores, even when detecting minority attack classes, in contrast to previous results for the CICIoV2024 dataset. The integration of optimized hyperparameter tuning and a broader methodological scope culminated in an IDS framework capable of addressing diverse attack scenarios with exceptional precision.

Data collection within a real-world environment may be compromised by several factors such as data-logger malfunctions and communication errors, during which no data is collected. As a consequence, appropriate tools are required to handle the missing values when analysing and processing such data. This problem is often tackled via matrix decomposition. While it has been successfully applied in a wide range of applications, in this work we report an issue that has been neglected in literature and “degenerates” the quality of the imputations obtained by matrix decomposition in multivariate time-series (with smooth evolution). Briefly, the problem consists of the misalignment of the matrix decomposition result: the missing values imputations fall within an incorrect range of values and the transitions between observed and imputed values are not smooth. We address this problem by proposing a post-processing alignment strategy. According to our experiments, the post-processing adjustment substantially improves the accuracy of the imputations (when the misalignment occurs). Moreover, the results also suggest that the misalignment occurs mostly when dealing with a small number of time-series due to lack of generalization ability.

Information security plays a key role in enterprises management, as it deals with the confidentiality, privacy, integrity, and availability of one of their most valuable resources: data and information. Small and Medium-sized enterprises (SME) are seen as a blind spot in information security and cybersecurity management, which is mainly due to their size, regional and familiar scope, and financial resources. This paper presents an information security and cybersecurity management project, in which a methodology based on the well-known ISO-27001:2013 standard was designed and implemented in fifty SMEs that were located in the center region of Portugal. The project was conducted by a business association located at the center of Portugal and mainly participated by SMEs. The Polytechnic of Leiria and an IT auditing/consulting team were the other two entities that participated on the project. The characterisation of the participating enterprises, the ISO-27001:2013 based methodology developed and implemented in SMEs, as well as the results obtained in this case study, are depicted and analysed in the paper. The attained results show a clear benefit to the audited and intervened SMEs, being mainly attested by the increasing of their information security management robustness and collaborators’ cyberawareness.

Information security and cybersecurity management play a key role in modern enterprises. There is a plethora of standards, frameworks, and tools, ISO 27000 and the NIST Cybersecurity Framework being two relevant families of international Information Security Management Standards (ISMSs). Globally, these standards are implemented by dedicated tools to collect and further analyze the information security auditing that is carried out in an enterprise. The overall goal of the auditing is to evaluate and mitigate the information security risk. The risk assessment is grounded by auditing processes, which examine and assess a list of predefined controls in a wide variety of subjects regarding cybersecurity and information security. For each control, a checklist of actions is applied and a set of corrective measures is proposed, in order to mitigate the flaws and to increase the level of compliance with the standard being used. The auditing process can apply different ISMSs in the same time frame. However, as these processes are time-consuming, involve on-site interventions, and imply specialized consulting teams, the methodology usually adopted by enterprises consists of applying a single ISMS and its existing tools and frameworks. This strategy brings overall less flexibility and diversity to the auditing process and, consequently, to the assessment results of the audited enterprise. In a broad sense, the auditing needs of Small and Medium-sized Enterprises (SMEs) are different from large companies and do not fit with all the existing ISMSs’ frameworks, that is a set of controls of a particular ISMS is not suitable to be applied in an auditing process, in an SME. In this paper, we propose a generic and client-centered web-integrated cybersecurity auditing information system. The proposed system can be widely used in a myriad of auditing processes, as it is flexible and it can load a set of predefined controls’ checklist assessment and their corresponding mitigation tasks’ list. It was designed to meet both SMEs’ and large enterprises’ requirements and stores auditing and intervention-related data in a relational database. The information system was tested within an ISO 27001:2013 information security auditing project, in which fifty SMEs participated. The overall architecture and design are depicted and the global results are detailed in this paper.

The next generation of mobile broadband communication, 5G, is seen as a driver for the industrial Internet of things (IIoT). The expected 5G-increased performance spanning across different indicators, flexibility to tailor the network to the needs of specific use cases, and the inherent security that offers guarantees both in terms of performance and data isolation have triggered the emergence of the concept of public network integrated non-public network (PNI-NPN) 5G networks. These networks might be a flexible alternative for the well-known (albeit mostly proprietary) Ethernet wired connections and protocols commonly used in the industry setting. With that in mind, this paper presents a practical implementation of IIoT over 5G composed of different infrastructure and application components. From the infrastructure perspective, the implementation includes a 5G Internet of things (IoT) end device that collects sensing data from shop floor assets and the surrounding environment and makes these data available over an industrial 5G Network. Application-wise, the implementation includes an intelligent assistant that consumes such data to generate valuable insights that allow for the sustainable operation of assets. These components have been tested and validated in a real shop floor environment at Bosch Termotecnologia (Bosch TT). Results show the potential of 5G as an enhancer of IIoT towards smarter, more sustainable, green, and environmentally friendly factories.

The adoption of Machine Learning Operations (MLOps) has grown rapidly as organisations seek to streamline the development and deployment of machine learning (ML) models. A core concept in MLOps workflows is the ML pipeline, consisting of a sequence of tasks representing the various stages of the ML lifecycle, such as data preprocessing, model training, and evaluation. As these tasks have different resource requirements and computational demands, using heterogeneous computing environments has become important. However, to exploit this heterogeneity, it is essential to map each task within a pipeline to the right machine. This paper introduces a modular and flexible placement system for ML pipelines that automatically allocates tasks to the most suitable machines in order to reduce execution and waiting times. Although designed to support custom placement strategies, the system employs a two-phase strategy: pipeline scheduling and task placement. During the scheduling phase, the Shortest Job First (SJF) algorithm determines the execution order of the pipelines. In the task placement phase, a heuristic-based method is used to assign tasks to machines. Experimental evaluations across a range of ML models and datasets demonstrate that the proposed system significantly outperforms baseline methods and the Kubernetes default scheduler. It achieved reductions of up to 68% in total execution time, and over 80% in average waiting time. Moreover, the system also demonstrates efficient pipeline dispatching in scenarios where multiple pipelines are submitted for execution. These results highlight the system’s potential to improve resource utilisation and accelerate ML model development in heterogeneous environments.

Digital exposure to the Internet among the younger generations, notwithstanding their digital abilities, has increased and raised the alarm regarding the need to intensify the education on cybersecurity in schools. Understanding of the human factor and its influence on children, namely their attitudes and behaviors online, is pivotal to reinforce their awareness towards cyberattacks, and to promote their digital citizenship. This paper aims to present an integrated cybersecurity and cyberawareness strategy composed of three major steps: (1) Cybersecurity attitude and behavior assessment, (2) self-diagnosis, and (3) teaching/learning activities. The following contributions are made: Two questionnaires to assess risky attitudes and behaviors regarding cybersecurity; a self-diagnosis to measure students’ skills on cybersecurity; a lesson plan addressing cyberawareness to be applied on Information and Communications Technology (ICT) and citizenship education curricular units. Cybersecurity risky attitudes and behaviors were evaluated in a junior high school population of 164 students attending the sixth and ninth grades. The assessment focused on two main subjects: To identify the attitudes and behaviors that raise the risk on cybersecurity among the participating students; to characterize the acquired students’ cybersecurity and cyberawareness skills. Global and individual scores and the histograms for attitudes and behaviors are presented. The items in which we have observed significant differences between sixth and ninth grades are depicted and quantified by their corresponding p-values obtained through the Mann–Whitney non-parametric test. Regarding the results obtained on the assessment of attitudes and behaviors, although positive, we observed that the attitudes and behaviors in ninth grade students are globally inferior compared to those attained by sixth grade students. The deployed strategy for cyberawareness was applied in a school context; however, the same approach is suitable to be applied in other types of organizations, namely enterprises, healthcare institutions and public sector.

Tampered multimedia content is being increasingly used in a broad range of cybercrime activities. The spread of fake news, misinformation, digital kidnapping, and ransomware-related crimes are amongst the most recurrent crimes in which manipulated digital photos and videos are the perpetrating and disseminating medium. Criminal investigation has been challenged in applying machine learning techniques to automatically distinguish between fake and genuine seized photos and videos. Despite the pertinent need for manual validation, easy-to-use platforms for digital forensics are essential to automate and facilitate the detection of tampered content and to help criminal investigators with their work. This paper presents a machine learning Support Vector Machines (SVM) based method to distinguish between genuine and fake multimedia files, namely digital photos and videos, which may indicate the presence of deepfake content. The method was implemented in Python and integrated as new modules in the widely used digital forensics application Autopsy. The implemented approach extracts a set of simple features resulting from the application of a Discrete Fourier Transform (DFT) to digital photos and video frames. The model was evaluated with a large dataset of classified multimedia files containing both legitimate and fake photos and frames extracted from videos. Regarding deepfake detection in videos, the Celeb-DFv1 dataset was used, featuring 590 original videos collected from YouTube, and covering different subjects. The results obtained with the 5-fold cross-validation outperformed those SVM-based methods documented in the literature, by achieving an average F1-score of 99.53%, 79.55%, and 89.10%, respectively for photos, videos, and a mixture of both types of content. A benchmark with state-of-the-art methods was also done, by comparing the proposed SVM method with deep learning approaches, namely Convolutional Neural Networks (CNN). Despite CNN having outperformed the proposed DFT-SVM compound method, the competitiveness of the results attained by DFT-SVM and the substantially reduced processing time make it appropriate to be implemented and embedded into Autopsy modules, by predicting the level of fakeness calculated for each analyzed multimedia file.