Agrivoltaics, the integration of photovoltaic (PV) systems with agricultural practices, presents a promising opportunity for sustainable energy and food production. This paper presents a case study aimed at closely examining the design considerations involved in integrating PV technology into greenhouses, with a focus on addressing the energy-food nexus and providing a potential solution to mitigate the competition between energy and food production. Through extensive data collection and analysis, valuable insights and guidance are obtained to contribute to the future development of design methodologies for agrivoltaic systems. Key findings from this case study reveal the influence of PV integration on microclimate parameters such as temperature and photosynthetically active radiation, as well as crop yield.
Motivation: This research addresses the critical need to identify drug-drug interactions (DDIs) before market entry. Existing preclinical detection methods are resource-intensive, prompting the use of computational models based on premarket drug properties. However, current models often oversimplify interactions, neglecting nuanced alterations in pharmacological effects. DDIs, rooted in the structural features of the DDI graph, are non-random, and understanding these relationships is vital for making comprehensive predictions and uncovering structural patterns in the DDI graph. This study introduces the Similarity Network Fusion and Convolutional Neural Networks (SNF-CNN) model, treating comprehensive DDIs as a signed network. Results: SNF-CNN excels in predicting degressive (AUC = 0.975, AUPR = 0.967), enhancive (AUC = 0.969, AUPR = 0.822) and Unknown DDIs (AUC = 0.971, AUPR = 0.948). A comparative analysis against state-of-the-art methods highlights the superiority of SNF-CNN, not only in predicting DDIs but also in accurately forecasting non-DDIs. The graphical abstract of SNF-CNN is provided (Figure 1). Availability and implementation: The SNF-CNN and data are available as open-source from GitHub at: https://github.com/aminkhod/SNF-CNN. For inquiries or collaboration, please contact A.khodamoradi@uninova.pt.
The main contribution of this work is based on the presentation of new AI models for the detection of attacks, within IoT systems using an extensive and complete dataset. In this context, we evaluate not only the performance of the models in terms of detection of attacks, but also their resource consumption, such as the time needed to analyze a sample, the consumption of computing cycles to analyze a sample, as well as the hard disk usage to store the AI models. Its application is oriented to the context of IoT systems in rural environments, where devices deployed in these environments usually have strong restrictions on these resources. Our results indicate that the OPTIMIST-LSTM model offers the best balance between accuracy and generalization, whereas XAI-IoT stands out for its computational efficiency, making them the most suitable for implementation in IoT infrastructures with limited resources.
Interoperability in Industrial Internet of Things (IIoT) environments remains a major challenge due to heterogeneous communication protocols and data models. Systems from distinct ecosystems spanning multiple layers, such as industrial fields, control, management, and cloud face interoperability issues at technical, structural, and semantic levels. Semantic interoperability is particularly complex, requiring syntactic and semantic alignment to ensure transmitted information is correctly understood. Current solutions rely on predefined agreements or tailored mechanisms, falling short of autonomous content translation. This paper describes a semantic gateway approach to enabling syntactic and semantic compliance between systems operating in distinct ecosystems, thereby supporting interoperability between heterogeneous providers and consumers. For that purpose, DITAG-Manager is proposed as a semantic gateway, coordinating the use of DITAG-Tool to generate translators and handling translation requests for exchanged content. A proof-of-concept (POC) was developed for experimental validation using the Spring Boot framework. The POC results show that the proposed approach facilitates straightforward, seamless, and more open service provisioning between heterogeneous systems, expressing soft real-time constraints compliance suitable for practical IIoT environments.
The widespread adoption of Internet of Things (IoT) technology in rural areas has led to qualitative leaps in fields such as agriculture, livestock farming, and transportation, giving rise to the concept of Smart Rural. However, Smart Rural IoT ecosystems are often vulnerable to cyberattacks. Although Artificial Intelligence (AI) based intrusion detection systems offer an effective solution to protect these environments, IoT devices are typically constrained in terms of memory and computation capabilities, making it essential to optimise the computational burden of AI models. This work explores different feature selection techniques to develop compact and fast Random Forest models for anomaly detection in IoT environments. The obtained results demonstrate that appropriate feature selection can reduce model size and inference time by at least 45 and 8
