This work presents the design and implementation of an IoT-based system that monitors the groundwater from the water table in a farming area in Argentina. Particularly, the system measures in real-time the fluctuations in the depth and quality of groundwater, and it represents a proof-of-concept led by the National Institute of Agriculture Technology (INTA). In this first stage, the system allows INTA to create a sandbox to both understand the behavior of groundwater and determine appropriate ways to deploy the monitoring system in semi-arid areas that produce crops. The management of soil water is mandatory to ensure the subsistence of life in the planet. It is particularly important in a climate change scenario, where water resources are becoming increasingly scarce. For that reason, several research projects are currently under development to understand the water table fluctuations, and based on it, to implement preparedness plans for particular areas. To the best of the authors knowledge, this is the first monitoring system of this kind that is being developed and deployed in the southern cone of Latin America. It is expected that the system allows farmers and government authorities to perform a more efficient soil water management and forecasting, reducing thus the uncertainty about the capability of particular areas to produce crops.
As driver assistance technologies become increasingly autonomous and pervasive, there is a growing demand for co-pilot systems that are not just technically sophisticated but also adaptable to individual drivers and dynamic situations. Modern Advanced Driver Assistance Systems (ADAS) fall short of offering personalized and context-sensitive support, leading to low user uptake and effectiveness. This article offers a conceptual review of new directions in intelligent driver assistance based on advances in human–machine interaction, behavior modeling, and affective computing. We outline the main shortcomings of current solutions and suggest a framework of design principles to shape the next generation of co-pilot systems. These principles are focused on continuous personalization, real-time contextual sensitivity, proactive assistance, and user-oriented interaction and are aimed at fostering a more empathic and adaptive driving experience. Our contribution is intended to guide the development of co-pilots that learn with users, improve safety and comfort, and fit into the vision of human-centered mobility systems, while also outlining directions for empirical validation and ethical deployment.
Technological advancements are opening new opportunities for managing chronic diseases such as diabetes. Current systems enable real-time glucose monitoring, but recent studies show that incorporating physiological and activity variables can improve glucose prediction. This work presents the architecture and implementation of an IoT infrastructure that combines glucose monitoring with physiological data from a sports wristband, together with contextual events such as carbohydrate intake and insulin administration. Glucose values are obtained through the FreeStyle Libre 2 sensor and the DiaBox app, while a custom Android application collects wristband data. All information is transmitted in real time via MQTT for structured storage. This infrastructure provides a unified dataset of glucose, physiological, and contextual variables, laying the foundation for the development of AI-based predictive systems to improve diabetes management.
There are several scenarios where the number and location of nodes in a network of meteorological stations may not be sufficient to obtain the information needed. An interesting alternative is the generation of virtual meteorological stations by applying a model that allows the limited information available to be used to estimate meteorological variables in areas where installing a physical station is not feasible. This paper presents a quantum computing-based model for generating virtual nodes in a network of meteorological stations. The properties of the generated virtual nodes are based on the estimation of meteorological variables from scarce or incomplete information from nearby nodes. The preliminary results of the software prototype that was built are presented. The quality of the estimate is assessed by comparing the figures produced by the model with control data from real historical sources.
This work presents an indoor localization system based on Ultra-Wide Band technology, enhanced with a semantic inference layer through graph-based modeling. These systems, focused on the person using them, have high applicability in the context of the Internet of Things and the Internet of Everything, as they provide not only precise location but also contextual information through reliable room identification. The proposed methodology models the monitored space as an undirected graph, where the nodes represent rooms defined by bounding boxes, and the edges indicate physically valid transitions. This improves spatial coherence and reduces erroneous room changes. The solution was validated in the SmartLab of the University of Jaén, achieving a 94.08
Ensuring privacy is becoming one of the biggest challenges for digital service providers in the era of the Internet of Everything (IoE), where billions of interconnected devices generate vast amounts of sensitive data. Metadata is one of the fundamental building blocks for developing Privacy-Enhancing Technologies (PETs), which can assist with the assessment of privacy issues. The existence of metadata repositories facilitates the development of these tools. However, effective exploitation is not limited to querying these repositories. It is essential to ensure that the available data is of a high enough quality to be trusted for validating applications, detecting conflicts, and proposing solutions to these issues. This thesis presents an innovative approach to this challenge by combining ontologies with artificial intelligence techniques. To the best of our knowledge, this has not been explored in this context before. As a proof of concept, the approach will be applied to the App-PIMD repository, which contains metadata from over 13,000 mobile applications.
Hazardous events such as floods, landslides, and other natural or anthropogenic disasters have posed significant challenges to public safety and territorial planning in Ecuador over the past decade. Understanding the spatial and temporal dynamics of such events is essential for effective risk management and the development of early warning systems. This paper presents a spatiotemporal analysis and predictive modeling framework for hazardous events in Ecuador, based on open government data from 2010 to 2023. Using a national-level dataset published by the Ecuadorian Secretariat for Risk Management, we explore the spatial, temporal, and categorical distribution of disaster-related incidents. We perform a comprehensive preprocessing pipeline, including temporal normalization, spatial discretization, and feature engineering. Our preliminary results highlight the viability of data-driven approaches in supporting early warning systems and guiding territorial planning efforts. The proposed methodology contributes to the development of more resilient and inclusive risk management strategies in Ecuador.
Within the realms of managing smart environments, especially within multi-user laboratory settings, predicting accurate occupancy faces a range of challenges. This study aims to address this challenge by utilising data gleaned from an IoT sensor network, and a digital twin paradigm. To enhance the accuracy of occupancy prediction Generative Adversarial Networks (GANs) has been implemented to generate synthetic data that can overcome the limitation of scarcity of the historical data that is gathered from the IoT sensor network. Long Short-Term Memory (LSTM) networks and Random Forest algorithms were used for the predictive modeling. In order to forecast occupancy rates, the models were trained on both synthetic datasets created using GANs and real datasets. The performances of the models found that the F1 score for the LSTM model was 0.801 without the synthetic data in the training and 0.802 with GAN for both the models. This strategy promotes more effective building management techniques in addition to optimizing the utilization of physical spaces.
This work presents a fully customized Bluetooth Low Energy (BLE) beacon system, developed using the Zephyr real-time operating system (RTOS). The beacon broadcasts structured, application-specific data through the Manufacturer Specific Data field, enabling a level of customization not offered by standard BLE profiles. A cross-platform mobile application, built with Flutter, has been designed to detect and interpret these customized signals in real time, providing a clear and interactive representation of the transmitted data. The complete system offers a lightweight, low-cost, and adaptable alternative to commercial solutions, particularly suitable for context-aware environments, smart infrastructures, and experimental research. The proposed architecture bridges embedded BLE development with mobile technologies, offering an open and scalable framework for customized wireless communication.
Transitional Interfaces (TIs) in Extended Reality (XR) enable users to fluidly transition between real and virtual environments, offering new opportunities to ease collaborative tasks in shared physical spaces such as libraries. In collaborative environments where users operate across the reality-virtuality continuum, differences in perceptual context can create barriers to effective communication. This challenge is particularly pronounced in co-located scenarios. To address this, we developed a cross-reality application for Meta Quest headsets that uses various attention markers in TIs to help users coordinate object-finding tasks. A user study in a library evaluated the effectiveness of four markers: a Basic Marker for simple highlighting, a Pathfinding Marker that indicates a route, a Gaze-Adaptive Pathfinding Marker that aligns guidance with the user’s gaze, and a Funnel Marker that broadly directs attention based on gaze direction. The results indicate that attention markers enhance collaborative efficiency, minimize focus shifts, and improve overall usability. Among the tested approaches, pathfinding markers yielded the best performance, while gaze-adaptive variants provided additional support in maintaining user focus. These findings suggest that XR-based attention guidance can effectively mitigate perceptual discrepancies in co-located collaborative tasks.
Brain tumor classification plays an important role in the early diagnosis and treatment of both cancerous and noncancerous tumors. Traditional techniques for Magnetic resonance imaging (MRI) scan analysis require a lot of time and can sometimes cause errors. In recent years, deep learning models have been used to speed up the process with fewer chances of errors. The study involves a transfer learning technique employing pretrained models including ResNet101, ResNet50, VGG16. The models are then trained on a diverse, labeled datasets of MRI scans obtained from various open-source websites. To enhance the efficacy of a model, a Gabor filter is applied to ResNet-101. The results revealed that the accuracy achieved by ResNet-101 with Gabor filters is 95
This work presents a simulation-based approach for contactless current estimation in three-phase segmented power cables using magnetic field sensing and data-driven modelling. The method aims to reconstruct symmetric phase currents from externally sampled magnetic field measurements at a realistic distance of 28.5 mm, replicating the sensor layout of a physical prototype. A high-fidelity finite element model is used as a digital twin of the real system, generating synthetic data under ideal load conditions. These data were used to assess current-estimation feasibility with lightweight, computationally efficient models. The results support the viability of non-invasive current sensing based on external magnetic field sampling and provide a baseline for future smart grid applications.
Virtual Humans (VHs), the AI-powered and photorealistic multimodal digital agents are capable of simulating human-to-human interactions. They are increasingly deployed across ubiquitous computing environments to support essential services in healthcare, education, and customer service. Yet, sustained user engagement remains a challenge. This study investigates how age influences user perceptions and adoption intentions of VHs across three dimensions: perceived wellbeing benefits, likelihood of use, and contextual usefulness. A cross-sectional online survey (N = 773) was conducted using a stratified age sample (18–80 years). Statistical analysis revealed significant associations between age and all three dimensions of VH perceptions, with young adults (18–30) and adults (31–50) consistently reporting higher belief in VH effectiveness and stronger intent to use. Older participants showed selective endorsement, often tied to specific use cases. These findings underscore the need for age-inclusive VH design that supports trust, usability, and relevance across ambient and assistive contexts. The study offers actionable insights for developing adaptive, user-centred VH systems that promote equitable access to wellbeing and quality of life enhancing interactive technologies. The paper discusses implications for ubiquitous human-agent interaction design and future research on age-aware digital agents. This study contributes to advancing inclusive human-computer interaction within Ambient, Active, and Assisted Living (A3L) environments.
Ambient Assisted Living (AAL) systems are becoming increasingly important for providing personalised assistance in smart homes. One key component for such systems is detecting and localising humans in different areas of the home, which can enhance contextual information to provide efficient support to the human user. Recent approaches often lack interpretability and compromise user privacy. This work introduces an interpretable, room-level human presence detection system that relies solely on low-cost, privacy-conserving ambient sensors typically used in smart homes. We have developed and evaluated a solution for presence detection based on data collected from a single participant in the Robot House, an ambient assisted living space at the University of Hertfordshire. We developed two models to perform this task, a Random Forest (RF) model and a more complex Long Short-Term Memory (LSTM) model across a triad of test scenarios, including full sensor set, sensor dropout and room dropout. We tested the performance of both models using conventional train-test splits and on an entirely unseen data to assess the generalisation. While LSTM achieved comparable results, RF performed better on new, unseen data, with an accuracy of 91.43
Accurate activity recognition in smart environments is vital for assessing an occupant’s ability to perform Activities of Daily Living (ADLs), indicating independent living. However, sensor errors like false readings and missing data often reduce system reliability. Traditional methods - Neural Networks, Dynamic Bayesian Networks, and Hidden Markov Models struggle with uncertainty in incomplete or unreliable sensor data. This work applies Dempster-Shafer (DS) theory to better handle such uncertainty, leveraging its reasoning mechanism to manage sensor inaccuracies. Using a dataset of nine representative ADLs from 32 activities captured via first-person video, with real-world imperfections like artifacts and frame loss, the DS-based system was evaluated against a traditional machine learning approach. Results show DS theory improves recognition performance under high uncertainty, demonstrating its potential for real-world smart environment applications.
Data visualization is a multidisciplinary domain that applies cognitive and perceptual principles to transform complex datasets into clear and understandable visual representations that guide decision-making. This work introduces a specialized data-visualization tool designed to support professionals and researchers in human gait and its underlying patterns. To collect data, non-invasive sensorized insoles are used, which provide valuable insights into the user’s movement without disrupting their natural gait. These insoles capture precise biomechanical data, including pressure distribution and movement dynamics, which are then analyzed through the tool interface. Such research contributes to the early detection of neurodegenerative disorders, where timely and accurate analysis of gait characteristics can enhance diagnostic precision and improve patient outcomes. The tool offers an intuitive interface and domain-specific analytical workflows that make it accessible to professionals and researchers without requiring extensive training in data analysis. Moreover, the integration of post-processed data visualization, including heatmaps and center of pressure animations, facilitates the identification of subtle gait abnormalities, which may otherwise go unnoticed in traditional clinical settings. By accelerating the progress of gait analysis research, the proposed tool has the potential to facilitate clinical advancements, ultimately leading to better monitoring, diagnosis, and treatment of neurodegenerative diseases.
The integration of blockchain technology into supply chain management has shown significant potential in enhancing transparency, traceability and security, particularly in industries such as pharmaceuticals and food. This paper explores the role of blockchain in object tagging systems, which link physical items to their digital records, providing a tamper-proof trail from production to sale. Despite the benefits of blockchain, the physical tagging process, particularly in preventing tampering or removal of tags, remains a critical vulnerability. The paper critically reviews existing solutions, including Radio Frequency Identification (RFID), Near Field Communication (NFC), and Quick Response (QR) code, and discusses their shortcomings in ensuring the integrity of high-value or sensitive goods. Additionally, the paper examines emerging technologies like IBM’s crypto anchors and presents a cost analysis of blockchain adoption for object tagging. Ultimately, the paper underscores the need for further research and innovation in physical tagging security to fully realize the potential of blockchain in supply chain management.
The rapid growth of consumer-oriented Internet of Things devices, encompassing smart home applications like energy management, entertainment, security, and environmental control, has created a complex ecosystem of smart devices. Platforms such as Home Assistant (HA) simplify device integration and control, enabling users to query and command all devices from a single hub. However, users still encounter important challenges and a steep learning curve to fluently interact with each device. This study explores the initial user experience of individuals adopting HA for the first time to accomplish certain tasks involving interaction with smart devices. We conducted a comparative observational study asking participants to perform increasingly complex tasks with HA and by using a voice assistant as an interface to HA. Finally, we asked participants to reflect and brainstorm about features they would like the system to exhibit so they could accomplish the tasks more easily. We report key usability findings regarding observed and self-reported challenges, positive experiences and potentially new voice interaction scenarios to simplify HA functionality. Our findings offer insights for designing intuitive and adaptable interfaces for smart home platforms.
This paper uses Chaos Game Representation (CGR) to visualize and characterize daily habits. Household monitoring data are analyzed with Dissimilarity Structural Similarity Index (DSSIM) and Earth Mover’s Distance (EMD). The method classifies days into three types—standard, apathy, and wandering—and detects specific habits within standard days, offering insights into older adults’ routines. Experiments with both generated and real datasets demonstrate the feasibility of accurately clustering day types and identifying habits.
Precise indoor positioning remains a key challenge in ambient intelligence and ubiquitous computing, particularly in health-related and assisted living environments. This work explores a novel perspective by utilising consumer wearables, specifically the Pixel Watch 3 with embedded Ultra-Wideband (UWB) capabilities, to enable private, peer-to-peer indoor localisation without relying on external infrastructure for indoor location. We present a hybrid UWB-based ranging system that integrates Pixel Watch 3 with Decawave DWM3001CDK development boards, aiming to evaluate the feasibility of real-time, short-range interaction, and contextual awareness under edge-computing approach. A custom application for the smartwatch was developed to manage UWB communication and synchronise it with the embedded anchors. The system was tested under controlled conditions to assess performance and battery consumption. The results show stable short-range measurements with high accuracy and a multi-hour battery life, confirming that consumer wearables can support high-precision UWB systems for decentralised, privacy-preserving applications in health, safety, and smart environments.
