Environmental monitoring is essential for smart agriculture, renewable energy assessment, and climate-aware farm management. However, deploying autonomous sensing platforms in rural environments remains challenging due to energy constraints, communication reliability, and real-time processing requirements. This paper presents a modular, solar-powered environmental monitoring platform integrating LTE-M communication and TinyML-enabled edge sensing. The proposed system adopts a dual-microcontroller architecture, combining an Arduino Nano 33 BLE for real-time sensor acquisition and edge processing with an Arduino MKR NB 1500 dedicated to low-power wide-area communication. The platform integrates temperature, humidity, atmospheric pressure, rainfall, wind, and light sensors within a scalable framework. Two monitoring stations were deployed in rural regions of Romania to evaluate communication robustness, sensing stability, and energy autonomy. Field results demonstrate reliable LTE-M connectivity (4,306 RSSI samples; mean -75.51 dBm) and strong agreement with a regional weather station, with mean deviations of \( -0.71^{\circ} \)C (temperature), \( 4.98\% \)(humidity), and a stable pressure offset of -9.58 hPa attributable to altitude differences. Despite a total system cost of €315, the platform achieves measurement performance comparable to professional meteorological stations while maintaining long-term solar-powered operation. The proposed architecture provides a scalable and cost-effective solution for distributed smart agriculture and environmental monitoring applications.

Accurate modeling of outdoor wireless propagation in dense urban environments is essential for smart city connectivity. Deterministic ray-tracing techniques provide high-fidelity multipath insight; however they suffer from high computational cost and limited scalability in large 3D environments. This work proposes a hybrid framework combining MATLAB-based (MATLAB 2024b 24.2.0.2773142, 64-bit, 22 October 2024) ray tracing and Machine Learning for scalable Wi-Fi 7 channel analysis. A large dataset is generated over a realistic university campus across multiple frequency bands, transmit powers, and reflection/diffraction configurations. Several regression models are evaluated, with emphasis on transformer-based architectures. The FT-Transformer achieves a Mean Absolute Error (MAE) of $3.49$ dB, RMSE of $5.36$ dB, and an $R^2$ of $99.63\%$ for validation, reducing computation time from months of simulation to seconds at inference. The framework enables accurate and efficient surrogate modeling for network planning and digital twin applications.

Accurate modeling of outdoor Wi-Fi propagation in dense urban environments is essential for smart city connectivity. Deterministic ray-tracing techniques provide high-fidelity insight into multipath propagation but suffer from high computational cost and limited scalability in large 3D environments. This work investigates a hybrid approach that combines MATLAB-based ray-tracing simulations with Machine Learning to enable scalable Wi-Fi~7 network analysis. A large dataset is generated over a realistic simulated university campus, covering multiple frequency bands (2.4, 5, and 6~GHz), transmit power levels, and ray-tracing configurations with reflections and diffractions. Several regression models are evaluated, with emphasis on transformer-based architectures. Results show that the FT-Transformer accurately approximates ray-tracing outputs while reducing inference time from months to minutes. The proposed framework enables fast surrogate modeling of radio propagation and supports network planning and digital twin applications.

Environmental monitoring is essential for smart agriculture, renewable energy assessment, and climate-aware farm management. However, deploying autonomous sensing platforms in rural environments remains challenging because of energy constraints, communication reliability, and real-time processing requirements. This paper presents a modular, solar-powered environmental monitoring platform integrating LTE-M communication and TinyML-enabled edge sensing. The proposed system adopts a dual-microcontroller architecture that combines an Arduino Nano 33 BLE for real-time sensor acquisition and edge processing with an Arduino MKR NB 1500 dedicated to low-power wide-area communication. The platform integrates temperature, humidity, atmospheric pressure, rainfall, wind, and light sensors within a scalable framework. Two monitoring stations were deployed in rural regions of Romania to evaluate communication robustness, sensing stability, and energy autonomy. Field results demonstrated reliable LTE-M connectivity (4306 received signal strength indicator [RSSI] samples; mean $-75.51$ dBm) and strong agreement with a regional weather station, with mean deviations of −0.71 °C (temperature), $4.98\%$ (humidity), and a stable pressure offset of $-9.58$ hPa attributable to altitude differences. Despite a total system cost of €315, the platform achieved measurement performance comparable to that of professional meteorological stations while maintaining long-term solar-powered operation. The proposed architecture provides a scalable and cost-effective solution for distributed smart agriculture and environmental monitoring applications.

The rapid growth of prosumer photovoltaic installations has introduced significant supply–demand imbalances in modern power grids, motivating the development of energy management systems that can coordinate distributed resources without sacrificing local control responsiveness. This paper presents qToggleEMS, a distributed architecture that combines cloud-resident receding-horizon planning with edge-resident bounded-override control for prosumer sites equipped with photovoltaic generation, battery storage, and grid interconnection. The contribution is positioned at the systems-engineering level: a documented partitioning of responsibilities between a cloud planner (forecasting, price-aware scheduling) and an edge controller (sub-second actuation, autonomous fallback) that preserves planning quality while remaining operational under cloud–edge disconnection. The cloud component, powerHub, is implemented as a set of microservices communicating via MQTT and TimescaleDB; the edge component runs qToggleOS on an ARM single-board computer and accesses inverters directly via Modbus RTU, bypassing manufacturer-provided cloud APIs. The system was deployed at a commercial prosumer site for approximately two months using the prosumer-oriented optimization strategy. Compared with a within-period counterfactual baseline (the cost the site would have incurred under its previous flat-tariff contract), monthly energy costs decreased by 14–15%. An analytical projection of the producer-oriented strategy using historical day-ahead prices from OPCOM PZU suggests a revenue uplift of approximately 23%, pending field validation.

Lumbar spine degeneration assessment from magnetic resonance imaging (MRI) involves multiple anatomical conditions, spinal levels, and ordered severity categories, making conventional accuracy alone insufficient for a complete evaluation. In this study, we present a Beyond-Accuracy framework for automated lumbar spine degeneration assessment using the RSNA 2024 Lumbar Spine Degenerative Classification dataset. Five independent Vision Transformer (ViT) pipelines were developed for spinal canal stenosis (SCS), left and right neural foraminal narrowing (LFN/RFN), and left and right subarticular stenosis (LSS/RSS). Localized MRI crops centered on annotated spinal-level coordinates were used to classify severity as Normal/Mild, Moderate, or Severe. For each condition, 24 ViT architecture and hyperparameter configurations were explored across ViT-B/16, ViT-L/16, ViT-B/32, and ViT-L/32 backbones, followed by repeated-run stability analysis. The selected condition-specific models achieved approximately 82–92% validation accuracy, with SCS reaching the highest performance at approximately 92%. Conventional measures, including accuracy, precision, recall, and F1-score, were complemented by Ordinal Accuracy (OA), Condition-Level Ordinal Accuracy (CLOA), and three Patient-Level Ordinal Accuracy (PLOA) formulations to characterize severity ordering at global, condition, and patient levels. The results show that exact classification accuracy alone does not fully capture model behavior in structured severity prediction. Combining condition-specific ViT pipelines with multi-level ordinal-aware evaluation provides a more comprehensive framework for lumbar degeneration assessment and supports the development of more interpretable MRI-based decision-support systems.