Given the increasing demand for enhanced power quality and higher efficiency in industrial electrical systems and modern grids, this paper presents a comparative analysis of three-level and five-level Neutral Point Clamped (NPC) multilevel converters controlled by Predictive Current Control (PCC) strategies. The study focuses on their operation when connected both to the grid and to a three-phase load, with the primary objective of reactive power compensation and power factor correction. The evaluation encompasses critical performance indicators such as dynamic response, Total Harmonic Distortion (THD), power factor, and both transient and steady-state behavior. Detailed simulations carried out in MATLAB/Simulink provide a quantitative and qualitative assessment of the two converter topologies. Results demonstrate that the five-level NPC topology delivers superior compensation capability, substantially improving the quality of the current injected into the grid. In particular, it achieves a significant reduction in THD and brings the power factor closer to unity when compared to its three-level counterpart. Nevertheless, these advantages are accompanied by increased complexity in terms of circuit design, semiconductor count, and control algorithm implementation. The findings of this work offer a comprehensive technical perspective that can serve as a guideline for selecting the most appropriate topology depending on the trade-off between power quality requirements and implementation challenges. Applications of these results extend to active power filters, Flexible AC Transmission Systems (FACTSs), and smart grids, where reliability, efficiency, and controllability are critical.

The growing adoption of wearable devices creates a critical need for robust and secure Internet of Things (IoT) solutions to manage biometric data streams. Current architectures often lack emphasis on seamless data capture, secure cloud storage and integrated dashboard visualization. This research addresses these gaps by investigating and evaluating an IoT framework leveraging lightweight communication and real-time visualization for improved healthcare monitoring. Drawing primarily on recent peer-reviewed journals and reputable conference proceedings, we evaluate an IoT architecture that securely integrates wearable biometric data into a cloud-based dashboard. The system utilizes encrypted advertising packets (e.g., AES-128-CCM) to broadcast biometric signals, eliminating the need for permanent device pairing and minimizing the energy consumption. These packets are captured by our prototype ESP32-based gateway node, decrypted and forwarded to a secure cloud environment that ensures persistent storage and accessibility. The cloud-based dashboard provides doctors and end-users with real-time insights and long term data tracking. Emphasis was placed on evaluating the system’s low latency performance, energy efficiency and data confidentiality. System evaluation demonstrates that encrypted advertising packets can securely transmit biometric signals, while drastically reducing energy consumption and latency. Advertising once per second reduces energy consumption by 50%, with further halving the sampling rate boosting savings up to 90%. Our architecture maintains robust data confidentiality and efficient storage, enabling effective cloud-driven visualization. This study validates the feasibility of encrypted advertising packets for secure, stable, scalable and efficient biometric IoT data acquisition, offering potential for advancements in remote healthcare monitoring and broader biomedical research environments.

Safeguarding microgrids with decentralized generation presents challenges due to the reciprocal power flow and fluctuations in renewable energy sources. Conventional protection systems often fail to adapt to these dynamic conditions, resulting in unreliable operation. This work proposes an innovative methodology for the automatic detection and classification of faults, using a Long Short-Term Memory (LSTM) neural network. The LSTM network was selected for its proven ability to process time series data, allowing it to capture the complex transient signatures of faults, which is crucial for accurate analysis. The research utilizes an extensive set of synchrophasor data (PMU) obtained from detailed simulations of a microgrid model in the MATLAB/Simulink environment. This dataset includes a variety of fault scenarios, including line-to-ground, line-to-line, and three-phase faults. To prepare the data, signal processing techniques from the Signal Processing Toolbox are applied to extract relevant features. Subsequently, an LSTM neural network is designed and trained using the Deep Learning Toolbox to classify fault types with high precision. The results demonstrate that the proposed approach achieves high accuracy and robustness in identifying different types of faults. The methodology contributes to the advancement of adaptive protection systems, offering an intelligent and effective alternative to traditional methods, and reinforces the security and resilience of modern microgrids.

Six-phase induction machines offer inherent fault tolerance capability, reduced torque ripple, and improved reliability compared to a classical three-phase configuration. However, they impose demanding requirements on current control strategies. This study presents a comparative evaluation of Sliding Mode Control (SMC) and the well-established Proportional Integral regulator with Pulse Width Modulation (PI+PWM), both modelled and implemented in MATLAB/Simulink using the same system and identical test condition profiles to ensure a fair comparison. The mentioned controllers are evaluated using speed reversal tests at ±500 rpm (8 kHz sampling) and ±1000 rpm (12 kHz sampling), and further assessed under ±50% variations in magnetising inductance to analyse robustness. The results show that PI+PWM achieves significantly lower steady-state current tracking error, with root mean square error (RMSE) typically below 0.02 A, compared to 0.05 – 0.08 A for SMC. In terms of current quality, SMC maintains a total harmonic distortion (THD) of approximately 1.3% at low speed and 1.05% at high speed. In comparison, PI+PWM consistently remains below 1.1% and 0.9%, respectively, demonstrating stable performance across both operating conditions. Consequently, PI+PWM emerges as a low-complexity and effective solution for industrial applications with limited computational resources. In contrast, SMC remains advantageous in scenarios requiring strong disturbance rejection and robustness to significant parameter variations.

1. Background: Breast cancer is one of the most fatal cancers among women around the globe. The chances of surviving this cancer increase with early tumor detection, which is necessary for effective treatment. Traditional diagnostic techniques are ineffective and take a long time to identify tumors, and the results may be accurate or inaccurate. Objective: The main objective of this project is to build an ML-based classification model that can help doctors to detect breast cancer early and more accurately. This project also aims to provide an interactive interface for easy accessibility for healthcare usage. Materials/Methods: For this study, twelve Machine Learning Classification Algorithms are implemented and tested: Logistic Regression, K-Nearest Neighbour (KNN), Support Vector Machine (SVM), Decision Tree, Random Forest, Gradient Boosting, XGBOOST, Naive Bayes, ADA Boosting, Light GBM, Cat Boost and Artificial Neural Network (ANN). The study used the Wisconsin Breast Cancer Dataset (WBCD) from the UCI ML Repository. It contains 569 patient samples and 30 features. This dataset possesses the following features: Radius, Texture, Area, Perimeter, Smoothness, Compactness, Concavity, and Fractional Dimension. The target variable is Diagnosis, which is categorized as Malignant vs Benign. Results: The fifteen models were analyzed, evaluated and compared using five performance metrics: Accuracy, Precision, Recall, F1-Score and AUC-ROC. Among all the evaluated models, the Artificial Neural Network (ANN) outperformed other models with an accuracy of 97.37.%, with 97% Precision, Recall and F1-Score. The AUC-ROC is nearly 99.61%, meaning that the model is able to differentiate between malignant and benign tumours.

Introduction. The growing demand for coffee and the current market situation highlight the need to investigate the impact of bean origin on beverage quality.

Aim. This study aims to evaluate the physicochemical and sensory characteristics of Arabica Red Bourbon beans cultivated in different regions.

Samples. Green coffee beans from the 2023 harvest were collected in Rwanda (R, altitude 1500 m), Colombia (Col, 1850 m), the Democratic Republic of the Congo (Con, 1800 m), and El Salvador (S, 1500 m).

Methods. Roasting was performed in a Kaffelogic Nano 7 roaster using an identical profile. Bean color was analyzed using a Lightstells CM-100 Plus Roast Analyzer. TDS and extraction yield (EXT) were measured with a DiFluid R2 Extract. Moisture content was determined with an EM 120-HR moisture analyzer, and water activity was assessed with a Novasina LabMaster aw. Sensory evaluation was carried out using the cupping method.

Results. The color of roasted beans ranged from 61.4 to 62.5, while the color of ground coffee was between 72.5 and 75.4. The moisture content was highest in the Col and R samples (3.4%) and lowest in the Con sample (3.1%). The greatest moisture loss during roasting was observed in S and R (13.4%). The water activity of green beans ranged from 0.50 to 0.56 and decreased to 0.18–0.30 after roasting. The extraction yield varied between 20.03% and 21.21%, while the TDS values ranged from 1.23% to 1.30%. The least acidic sample was S (pH 5.04). An unusually high caffeine content was detected in the Colombian sample.

Conclusion. The conducted research confirmed that the geographical origin of Arabica Red Bourbon beans has a significant impact on their physicochemical and sensory attributes. Variations in moisture, acidity, and caffeine content were observed among the samples, despite a consistent roasting profile.

The aim of this study was to investigate the substitution of egg yolk in mayonnaise-type sauces with alternative protein components and to optimize the hydrocolloid composition for improved stability and rheological properties. Mustard powder (1 %), soybean flour (1 %), casein (2 %) and cream powder (1 %) blends were employed as emulsifiers. The influence of the ratio of potato starch, carboxymethylcellulose (CMC), pectin, and xanthan gum (0–1% each) on the properties of low-fat mayonnaise formulations (30% oil content) was examined.

The following methods were used for analysis: microscopy (Micromed microscope with an integrated 1k Pixelink camera), laser diffraction (PCA-1190), viscometry (Visco QC-300), and potentiometry (pH-150MI).

Sedimentation and thermal stability tests revealed high resistance of all samples (98-99%), both after 24 h and following 20–30 days of storage. Optical microscopy confirmed the homogeneity of the structure, with individual dispersed particles of 100–150 μm corresponding to inclusions of plant protein additives. The particle size distribution D [4,3] exhibited a bimodal profile, with peaked at 0.1–1 μm and 2–8 μm, indicating efficient homogenization of the emulsions. Storage experiments (near 30 days) demonstrated an increase in particle size by 1.4–1.6 times and a decrease in viscosity, likely due to flocculation and aggregation of polysaccharide clusters into larger agglomerates.

Among the tested formulations, the sample containing 0.3% CMC, 0.3% xanthan gum, and 0.4% pectin showed the most favorable physicochemical and sensory properties, highlighting the synergistic effect of hydrocolloid blends in stabilizing reduced-fat mayonnaise-type emulsions.

The growing reliance on renewable energy sources, particularly solar photovoltaics (PV), requires intelligent management strategies to address challenges of intermittency, storage, and efficiency in autonomous microgrids. This review investigates IoT-based solutions for energy optimization, focusing on hardware platforms, communication protocols, and intelligent control strategies that enhance the reliability and autonomy of PV-powered systems. We conducted a structured review of IoT-enabled solar microgrid applications, relying on peer-reviewed journal articles, reputable conference proceedings, and scholarly works published between 2020 and 2025. The focus centers on microcontroller-based platforms (e.g., Arduino, ESP32, NodeMCU, TTGO LoRa32) and Single Board Computers (SBCs) (e.g., Raspberry Pi), alongside the integration of optimization algorithms with Machine Learning (ML) and Neural Network (NN) approaches. Our results highlight that lightweight microcontrollers offer cost-effective monitoring, ESP32 and NodeMCU balance real-time analytics with energy efficiency, Raspberry Pi supports edge-level AI processing, and LoRa enables scalable long-range communication for remote PV systems. Furthermore, optimization algorithms (PSO, WOA-SA) and neural models (ANN, LSTM, CNN-LSTM) are explored as methods to improve forecasting accuracy, fault detection, and demand-side management. The conclusions indicate that IoT-based architectures significantly improve energy efficiency, support predictive maintenance, and enable scalable deployment of autonomous solar microgrids. The study underlines the necessity of hybrid IoT architectures, combining edge and cloud intelligence, to balance computational complexity, power constraints, and cybersecurity requirements. These findings provide practical insights into designing robust, cost-effective, and scalable IoT-enabled PV microgrids that contribute to decentralized and sustainable energy transitions.

The rapid expansion of the Internet of Things (IoT) has introduced a diverse set of devices operating in constrained environments, raising critical security concerns in domains such as smart homes, industrial automation, and healthcare. Many IoT ecosystems use lightweight wireless protocols for low-power, short-range communication. While these protocols embed security mechanisms, their alignment with formal cybersecurity assurance frameworks remains insufficiently studied. Drawing primarily on recent peer-reviewed journals and reputable conference proceedings, we evaluate Thread, Zigbee and Z-Wave against the Common Criteria (CC) Functional Requirements for Cryptography (FCS), as defined in CC:2022 and the European Union Cybersecurity Certification Scheme (EUCC). The assessment focuses on key CC components, including cryptographic key generation (FCS_CKM.1), distribution (FCS_CKM.2), agreement (FCS_CKM_EXT.7), operations (FCS_COP.1), and random bit generation (FCS_RBG.1). Our findings show that Thread demonstrates the strongest alignment with CC requirements, leveraging AES-CCM authenticated encryption and ECDH-based key exchange within a flexible, decentralized trust model. Zigbee provides comparable cryptographic strength but its reliance on a centralized Trust Center complicates compliance with key management lifecycle controls. Z-Wave has improved with the S2 Security framework, adopting ECDH exchanges, but still faces challenges due to proprietary constraints and limited transparency. This comparative analysis highlights that while all three protocols provide baseline security, only Thread is aligned with CC and EUCC certification schemes. Achieving compliance for Zigbee and Z-Wave will require protocol hardening and stricter cryptographic key lifecycle management. Aligning IoT protocols with CC is essential for building trust and resilience in critical connected systems.

This study evaluates the application of deep learning methods to the design of a microwave transmitter–receiver system operating in the mid-band of 5G communications. The proposed system comprises four stages—signal generation, amplification, mixing, and filtering—each designed individually using traditional microwave theory and then integrated into a full transceiver. Simulation data were generated in MATLAB and ADS, and four convolutional neural networks (CNNs) were implemented in Python (TensorFlow/Keras), with architectures ranging from 11 to 271 layers and training datasets between 4,000 and 12,000 samples. Training was performed over 200–1,000 epochs using Adam optimization, ReLU/linear activations, and sequential dense connections. Across all networks, the average error reduction exceeded 90%, with convergence achieved after the third training cycle for most components. For the transceiver integration, baseline design simulations indicated a transmitted power of –32.637 dBm with a gain of 1.116 dB. The deep learning-based design yielded comparable results, with a transmitted power of –33.912 dBm and a gain of 0.738 dB. These results demonstrate that the neural network models successfully approximated the behavior of microwave components without degrading system-level performance. Further analysis of scattering parameters (S-parameters) confirmed that the CNN-trained models maintained acceptable matching and frequency response across the 3.5 GHz operating band. Overall, this study demonstrates a complementary design methodology for microwave systems in 5G applications, enabling the modeling and optimization of multiple components simultaneously.