Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
7.0
2.9
Journals
Electronics
Editor’s Choice
electronics-logo

Journal Menu

► Journal Menu

Journal Browser

► Journal Browser
share announcement

Editor’s Choice Articles

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Order results
Result details
Results per page
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
35 pages, 2974 KB  
Article
Multi-Agent Coordination Strategies vs. Retrieval-Augmented Generation in LLMs: A Comparative Evaluation
by Irina Radeva, Ivan Popchev, Lyubka Doukovska and Miroslava Dimitrova
Electronics 2025, 14(24), 4883; https://doi.org/10.3390/electronics14244883 - 11 Dec 2025
Cited by 1 | Viewed by 3965
Abstract
This paper evaluates multi-agent coordination strategies against single-agent retrieval-augmented generation (RAG) for open-source language models. Four coordination strategies (collaborative, sequential, competitive, hierarchical) were tested across Mistral 7B, Llama 3.1 8B, and Granite 3.2 8B using 100 domain-specific question–answer pairs (3100 total evaluations). Performance [...] Read more.
This paper evaluates multi-agent coordination strategies against single-agent retrieval-augmented generation (RAG) for open-source language models. Four coordination strategies (collaborative, sequential, competitive, hierarchical) were tested across Mistral 7B, Llama 3.1 8B, and Granite 3.2 8B using 100 domain-specific question–answer pairs (3100 total evaluations). Performance was assessed using Composite Performance Score (CPS) and Threshold-aware CPS (T-CPS), aggregating nine metrics spanning lexical, semantic, and linguistic dimensions. Under the tested conditions, all 28 multi-agent configurations showed degradation relative to single-agent baselines, ranging from −4.4% to −35.3%. Coordination overhead was identified as a primary contributing factor. Llama 3.1 8B tolerated Sequential and Hierarchical coordination with minimal degradation (−4.9% to −5.3%). Mistral 7B with shared context retrieval achieved comparable results. Granite 3.2 8B showed degradation of 14–35% across all strategies. Collaborative coordination exhibited the largest degradation across all models. Study limitations include evaluation on a single domain (agriculture), use of 7–8B parameter models, and homogeneous agent architectures. These findings suggest that single-agent RAG may be preferable for factual question-answering tasks in local deployment scenarios with computational constraints. Future research should explore larger models, heterogeneous agent teams, role-specific prompting, and advanced consensus mechanisms. Full article
(This article belongs to the Special Issue Machine Learning Applications in Computer Vision, Data Modeling, and Natural Language Processing)
►▼ Show Figures

Figure 1

18 pages, 2727 KB  
Article
Heterogeneous Graph Neural Network for WiFi RSSI-Based Indoor Floor Classification
by Houjin Lu and Seung-Hoon Hwang
Electronics 2025, 14(24), 4845; https://doi.org/10.3390/electronics14244845 - 9 Dec 2025
Cited by 2 | Viewed by 1000
Abstract
Accurate indoor floor classification is essential for wireless positioning systems. However, the performance of conventional received signal strength indictor (RSSI)-based fingerprinting approaches is often limited by signal fluctuations and insufficient feature representation. To address these challenges, this paper introduces a heterogeneous graph neural [...] Read more.
Accurate indoor floor classification is essential for wireless positioning systems. However, the performance of conventional received signal strength indictor (RSSI)-based fingerprinting approaches is often limited by signal fluctuations and insufficient feature representation. To address these challenges, this paper introduces a heterogeneous graph neural network (GNN) framework that models WiFi signals using two types of nodes: reference points and Media Access Control (MAC) address. The edges between reference points and MAC addresses are weighted by normalized RSSI values, allowing the model to capture signal strength interactions and perform relation-aware message passing. Through this graph-based representation, the model can learn spatial and signal dependencies more effectively than conventional vector-based approaches. The proposed model was extensively evaluated under both benchmark and practical settings. On small-scale datasets, it achieved performance comparable to that of a conventional convolutional neural network trained on large-scale datasets, confirming its effectiveness with limited samples. In addition, the proposed model consistently outperformed other models under noisy conditions, achieving 93.88% accuracy on the widely used UJIIndoorLoc dataset and 97.3% accuracy in real-time experiments conducted at a test site. These values are significantly higher than those achieved using conventional machine learning (ML) baselines, highlighting the ability of the proposed model to handle real-world signal variations. These findings highlight that the heterogeneous GNN effectively captures spatial and signal-level dependencies, offering a robust and scalable solution for accurate indoor floor classification. Overall, this work presents a promising pathway for improving the performance and reliability of future wireless positioning systems. Full article
(This article belongs to the Special Issue Advanced Indoor Localization Technologies: From Theory to Application)
►▼ Show Figures

Figure 1

16 pages, 1309 KB  
Article
Ant Colony Optimization for CMOS Physical Design: Reducing Layout Area and Improving Aspect Ratio in VLSI Circuits
by Arnab A. Purkayastha, Jay Tharwani and Shobhit Aggarwal
Electronics 2025, 14(24), 4825; https://doi.org/10.3390/electronics14244825 - 8 Dec 2025
Viewed by 997
Abstract
This paper presents an enhanced Ant Colony Optimization (ACO) algorithm tailored for optimizing CMOS physical design in VLSI circuits. As device complexity escalates, traditional placement techniques struggle with multiobjective goals such as minimizing layout area, wirelength, and achieving effective aspect ratios. The proposed [...] Read more.
This paper presents an enhanced Ant Colony Optimization (ACO) algorithm tailored for optimizing CMOS physical design in VLSI circuits. As device complexity escalates, traditional placement techniques struggle with multiobjective goals such as minimizing layout area, wirelength, and achieving effective aspect ratios. The proposed ACO framework simulates artificial ant colonies exploring layout configurations and reinforcing promising solutions through a pheromone-guided heuristic. Evaluated on a benchmark containing ten typical logic blocks—Adder, Multiplier, Shifter, MUX, Register, ALU, Decoder, Control, Cache, and Buffer—the ACO method achieves a maximum layout area reduction of 27.27% (from 1760 to 1280 units2) and improves the aspect ratio from 3.64 to 5.0 compared to traditional layouts. The mean area reduction observed across different parameter settings is approximately 20%. The system also includes a fully configurable and modular automation tool designed for flexible parameter tuning and the rapid benchmarking of the ACO algorithm. This tool enables users to easily adjust key parameters such as number of ants, iteration count, pheromone evaporation rate, and heuristic influences, allowing for a comprehensive exploration of the optimization space. Experimental results demonstrate ACO’s scalability, adaptability, and effectiveness, establishing it as a viable approach for automation in complex physical designs. Future work will focus on hybrid algorithms and multi-objective optimization extensions. Full article
(This article belongs to the Special Issue Recent Advances in AI Hardware Design)
►▼ Show Figures

Figure 1

20 pages, 8243 KB  
Article
Evolution of Programmable Boards as a Response to Market Changes and Technological Paradigm Developments
by Emilian Świtalski and Krzysztof Górecki
Electronics 2025, 14(24), 4810; https://doi.org/10.3390/electronics14244810 - 6 Dec 2025
Cited by 1 | Viewed by 1056
Abstract
This article examines the evolution of programmable development boards in the context of dynamic market and technological changes, as well as new approaches to embedded system design. It presents the history of evaluation board development and its impact on engineering, emphasizing the shift [...] Read more.
This article examines the evolution of programmable development boards in the context of dynamic market and technological changes, as well as new approaches to embedded system design. It presents the history of evaluation board development and its impact on engineering, emphasizing the shift from closed ecosystems to open-source software. Particular attention is paid to the role of open source and modular architectures in shaping modern solutions. An important component of this paper is the description of the original OpenCPLC evaluation system, dedicated to microcomputer control systems. The OpenCPLC project illustrates how a layered design approach can increase the flexibility and range of PLC applications in both industry and education. The evolution of development boards indicates a clear trend in which openness and modularity are becoming key factors in determining the future of embedded systems. Full article
(This article belongs to the Section Industrial Electronics)
►▼ Show Figures

Figure 1

34 pages, 3902 KB  
Article
Comparing Explainable AI Models: SHAP, LIME, and Their Role in Electric Field Strength Prediction over Urban Areas
by Ioannis Givisis, Dimitris Kalatzis, Christos Christakis and Yiannis Kiouvrekis
Electronics 2025, 14(23), 4766; https://doi.org/10.3390/electronics14234766 - 4 Dec 2025
Cited by 18 | Viewed by 8621
Abstract
This study presents a comparative evaluation of state-of-the-art Machine Learning (ML) and Explainable Artificial Intelligence (XAI) methods, specifically SHAP and LIME, for predicting electromagnetic field (EMF) strength in urban environments. Using more than 19,000 in situ EMF measurements across Catalonia, Spain, combined with [...] Read more.
This study presents a comparative evaluation of state-of-the-art Machine Learning (ML) and Explainable Artificial Intelligence (XAI) methods, specifically SHAP and LIME, for predicting electromagnetic field (EMF) strength in urban environments. Using more than 19,000 in situ EMF measurements across Catalonia, Spain, combined with high-resolution geospatial features such as building height, built-up volume, and population density, six ML algorithms were trained and assessed over 50 randomized train–test splits. The k-Nearest Neighbors (kNN) model achieved the highest predictive accuracy (RMSE = 0.623), followed by XGBoost (RMSE = 0.711) and LightGBM (RMSE = 0.717). Explainability analysis showed that SHAP consistently identified built-up volume, building height, degree of urbanization, and population density as the dominant global predictors of EMF strength, whereas LIME revealed that degree of urbanization, population density, and building height were the most influential at the local (micro-scale) level. The results demonstrate that integrating interpretable ML frameworks with enriched geospatial datasets improves both predictive performance and transparency in EMF exposure modeling, supporting data-driven urban planning and public health assessment. Full article
►▼ Show Figures

Figure 1

26 pages, 736 KB  
Article
Communication-Efficient Federated Optimization with Gradient Clipping and Attention Aggregation for Data Analytics and Prediction
by Shengyuan Tang, Linwan Zhang, Shengzhe Xu, Xinyue Zeng, Peng Hu, Xinyi Gong and Manzhou Li
Electronics 2025, 14(23), 4778; https://doi.org/10.3390/electronics14234778 - 4 Dec 2025
Viewed by 1236
Abstract
To address the challenge of collaborative strategy optimization caused by non-independent and identically distributed data in cross-institutional scenarios, a federated quantitative learning framework integrating Path Attention Aggregation Module (PAAM), Gradient Clipping and Compression (GCC), and a Heterogeneity-Aware Adaptive Optimizer (HAAO) is proposed to [...] Read more.
To address the challenge of collaborative strategy optimization caused by non-independent and identically distributed data in cross-institutional scenarios, a federated quantitative learning framework integrating Path Attention Aggregation Module (PAAM), Gradient Clipping and Compression (GCC), and a Heterogeneity-Aware Adaptive Optimizer (HAAO) is proposed to achieve efficient return optimization and robust risk control. The framework is validated across multi-market and multi-institutional environments, with experiments covering three key dimensions: return performance, risk management, and communication efficiency. The results demonstrate that the proposed model achieves an annualized return (AR) of 16.57%, representing an approximate 19.7% improvement over the traditional FedAvg model; the Sharpe ratio (SR) increases to 1.25, while the maximum drawdown (MDD) decreases to 15.92% and volatility remains controlled at 8.83%, indicating superior balance between return and risk. In the communication efficiency evaluation, when the number of communication rounds is reduced to 50 and 25, the model maintains accuracy at 94.2% and 92.8%, recall at 93.5% and 91.7%, and precision at 94.8% and 92.3%, respectively. Overall, the proposed framework achieves a dynamic balance between global convergence and risk constraints through path weighting, gradient sparsification, and frequency-domain learning rate adjustment. This research provides a novel and scalable paradigm for distributed financial prediction that ensures both privacy preservation and communication efficiency, demonstrating substantial engineering feasibility and practical applicability in intelligent financial modeling. Full article
(This article belongs to the Special Issue Machine Learning in Data Analytics and Prediction)
►▼ Show Figures

Figure 1

9 pages, 1376 KB  
Article
Comparative Study on Device Type Configurations of 2T0C DRAM for Compute-in-Memory Applications
by Seonghwan Kong and Wonbo Shim
Electronics 2025, 14(23), 4742; https://doi.org/10.3390/electronics14234742 - 2 Dec 2025
Cited by 2 | Viewed by 1765
Abstract
In this article, we systematically analyze the electrical characteristics of 2T0C DRAM unit cells for compute-in-memory (CIM) applications, focusing on the on/off current ratio, coupling effects, and retention time, with respect to the NN (NMOS-NMOS), NP (NMOS-PMOS), PN (PMOS-NMOS), and PP (PMOS-PMOS) device [...] Read more.
In this article, we systematically analyze the electrical characteristics of 2T0C DRAM unit cells for compute-in-memory (CIM) applications, focusing on the on/off current ratio, coupling effects, and retention time, with respect to the NN (NMOS-NMOS), NP (NMOS-PMOS), PN (PMOS-NMOS), and PP (PMOS-PMOS) device types. We designed 65 nm CMOS-based 2T0C DRAM unit cells using Sentaurus 3D TCAD simulation, while keeping the physical parameters and doping concentrations of the NMOS and PMOS transistors identical to impartially compare their electrical characteristics. As a result, the NP- and PN-type 2T0C DRAM cells exhibited much shorter retention times for data ‘0’ (<0.55 ns) and smaller on/off current ratios (<4.36) than the NN- and PP-type cells, clearly indicating that the NP- and PN-type cells are not favorable for CIM applications. Furthermore, we investigated the cause of the storage node voltage fluctuation immediately after a read pulse using the equivalent RC models of the NN- and PP-type cells. The origin of this fluctuation is attributed to three factors: the coupling effects between the storage node and the read transistor, the drain current in the read transistor, and the subthreshold leakage current in the write transistor. Full article
(This article belongs to the Section Semiconductor Devices)
►▼ Show Figures

Figure 1

19 pages, 2731 KB  
Article
Adaptive Channel-Aware Garbage Collection Control for Multi-Channel SSDs
by Hyunho Mun and Youpyo Hong
Electronics 2025, 14(23), 4741; https://doi.org/10.3390/electronics14234741 - 2 Dec 2025
Viewed by 900
Abstract
Solid-State Drives (SSDs) have become the dominant storage medium in performance-sensitive systems due to their high throughput, reliability, and energy efficiency. However, inherent constraints in NAND flash memory—such as out-of-place writes, block-level erase operations, and data fragmentation—necessitate frequent garbage collection (GC), which can [...] Read more.
Solid-State Drives (SSDs) have become the dominant storage medium in performance-sensitive systems due to their high throughput, reliability, and energy efficiency. However, inherent constraints in NAND flash memory—such as out-of-place writes, block-level erase operations, and data fragmentation—necessitate frequent garbage collection (GC), which can significantly degrade user I/O performance when not properly managed. This paper presents a channel-aware GC control mechanism for multi-channel SSD architectures that limits GC concurrency based on real-time storage utilization. Unlike conventional controllers that allow GC to proceed simultaneously across all channels—often leading to complete I/O stalls—our approach adaptively throttles the number of GC-active channels to preserve user responsiveness. The control logic uses a dynamic thresholding function that increases GC aggressiveness only as the SSD approaches full capacity, allowing the system to balance space reclamation with quality-of-service guarantees. We implement the proposed mechanism in an SSD simulator and evaluate its performance under a range of real-world workloads. Experimental results show that the proposed adaptive GC control significantly improves SSD responsiveness across various workloads. Across all workloads, the proposed adaptive GC control achieved an average latency improvement factor of 4.86×, demonstrating its effectiveness in mitigating GC-induced interference. Even when excluding extreme outlier cases, the method maintained an average improvement of 1.55×, with a standard deviation of 1.17, confirming its consistency and robustness across diverse workload patterns. Full article
(This article belongs to the Section Computer Science & Engineering)
►▼ Show Figures

Figure 1

35 pages, 3408 KB  
Article
Unleashing GHOST: An LLM-Powered Framework for Automated Hardware Trojan Design
by Md Omar Faruque, Peter Jamieson, Ahmad Patooghy and Abdel-Hameed A. Badawy
Electronics 2025, 14(23), 4745; https://doi.org/10.3390/electronics14234745 - 2 Dec 2025
Cited by 7 | Viewed by 2386
Abstract
Traditionally, inserting realistic Hardware Trojans (HTs) in complex hardware systems has been a time-consuming manual process, requiring comprehensive knowledge of the design and navigating intricate Hardware Description Language (HDL) codebases. Machine Learning (ML)-based approaches have attempted to automate this process but often struggle [...] Read more.
Traditionally, inserting realistic Hardware Trojans (HTs) in complex hardware systems has been a time-consuming manual process, requiring comprehensive knowledge of the design and navigating intricate Hardware Description Language (HDL) codebases. Machine Learning (ML)-based approaches have attempted to automate this process but often struggle with the need for extensive training data, learning time, and limited generalizability across diverse hardware design landscapes. This paper introduces GHOST, an automated tool that leverages Large Language Models (LLMs) for rapid generation and insertion of HT. The research encompasses both the development of the GHOST framework and a comprehensive evaluation of its effectiveness across three state-of-the-art LLMs-GPT-4, Gemini-1.5-Pro, and Llama-3-70B. According to our evaluations, GPT-4 demonstrates the best performance by successfully generating and inserting HTs in 88.9% of its attempts. This study also highlights the security risks posed by LLM-generated HTs, as 100% of successful GHOST-generated HTs that completed inference within the time limit evaded detection by a state-of-the-art ML-based HT detection tool. These results underscore the need for advanced detection and prevention mechanisms in hardware security to address the emerging threat of LLM-generated HTs. Full article
(This article belongs to the Section Artificial Intelligence)
►▼ Show Figures

Figure 1

17 pages, 25231 KB  
Article
Low-Cost and Fully Metallic Reconfigurable Leaky-Wave Antenna Based on 3D-Printing Technology for Multi-Beam Operation
by Miguel Díaz-Martín, Carlos Molero, Ginés Martínez-García and Marcos Baena-Molina
Electronics 2025, 14(23), 4723; https://doi.org/10.3390/electronics14234723 - 30 Nov 2025
Viewed by 1117
Abstract
Global data consumption is experiencing exponential growth, driving the demand for wireless links with higher transmission speeds, lower latency, and support for emerging applications such as 6G. A promising approach to address these requirements is the use of higher-frequency bands, which in turn [...] Read more.
Global data consumption is experiencing exponential growth, driving the demand for wireless links with higher transmission speeds, lower latency, and support for emerging applications such as 6G. A promising approach to address these requirements is the use of higher-frequency bands, which in turn necessitates the development of advanced antenna systems. This work presents the design and experimental validation of a reconfigurable, low-cost leaky-wave antenna capable of controlling the propagation direction of single-, dual-, and triple-beam configurations in the FR3 frequency band. The antenna employs slotted periodic patterns to enable directional electromagnetic field leakage, and it is based on a cost-effective and simple 3D-printing fabrication process. Laboratory testing confirms the theoretical and simulated predictions, demonstrating the feasibility of the proposed antenna solution. Full article
(This article belongs to the Special Issue Recent Advancements in Antenna Technologies: Novel Design Techniques and Emerging Applications)
►▼ Show Figures

Figure 1

28 pages, 3446 KB  
Article
Reaction Wheel Pendulum Stabilization Using Various State-Space Representations
by Jacek Michalski, Mikołaj Mrotek, Tymoteusz Tomczak, Jakub Wojciechowski and Dariusz Pazderski
Electronics 2025, 14(23), 4719; https://doi.org/10.3390/electronics14234719 - 29 Nov 2025
Cited by 1 | Viewed by 1468
Abstract
This paper addresses the problem of stabilizing an inverted pendulum actuated by a reaction wheel, a system relevant for robotic balancing platforms and aerospace applications. The study compares several state-space representations of the system and examines their implications for controller synthesis and parameter [...] Read more.
This paper addresses the problem of stabilizing an inverted pendulum actuated by a reaction wheel, a system relevant for robotic balancing platforms and aerospace applications. The study compares several state-space representations of the system and examines their implications for controller synthesis and parameter identification. A unified nonlinear model formulation is introduced, enabling a structural Lyapunov-based robustness analysis that reveals how variations in the gravitational gain affect closed-loop stability. Control strategies based on pole placement and Linear Quadratic Regulator (LQR) design are implemented and compared across the different representations. The analysis highlights a robustness–fidelity trade-off between model complexity and sensitivity to parameter uncertainty, providing insight that extends beyond the specific laboratory setup. Theoretical results are validated on a real laboratory platform. The controllers are evaluated in both upright and downward equilibrium configurations, and the influence of parameter shifts is assessed experimentally using global identification and performance indices. The work offers general modeling and robustness guidelines for reaction-wheel-based stabilization systems and related underactuated nonlinear mechanisms. Full article
(This article belongs to the Section Systems & Control Engineering)
►▼ Show Figures

Figure 1

17 pages, 4978 KB  
Article
Optimizing Periodic Intervals in Multi-Stage Waveguide Stub Bandstop Filters for Microwave Leakage Suppression
by Yusuke Kusama, Hao-Hui Chen, Yao-Wen Hsu, Kyohei Murayama and Robert Weston Johnston
Electronics 2025, 14(23), 4660; https://doi.org/10.3390/electronics14234660 - 27 Nov 2025
Cited by 1 | Viewed by 856
Abstract
Waveguide bandstop filters (BSFs) play a key role in preventing electromagnetic wave leakage from gaps or sample entrances and exits, which can compromise safety, work efficiency, and electromagnetic compatibility. This study designs a waveguide BSF using a finite periodic structure of cascaded short-circuited [...] Read more.
Waveguide bandstop filters (BSFs) play a key role in preventing electromagnetic wave leakage from gaps or sample entrances and exits, which can compromise safety, work efficiency, and electromagnetic compatibility. This study designs a waveguide BSF using a finite periodic structure of cascaded short-circuited E-plane stubs (chokes) to achieve a stopband with the transmission coefficient |S21| ≤ −30 dB and a 4% relative bandwidth. We investigate the impact of stub width on bandwidth broadening and stub spacing in cascade connections on spurious passband suppression. Electromagnetic and circuit simulations, validated experimentally, reveal that stub spacing at odd multiples of a quarter guided wavelength (λg/4) minimizes spurious passbands, with wider stubs and larger spacings enhancing stopband characteristics. This indicates that there is a great advantage in reducing the number of resonators and manufacturing costs. These findings provide practical and new design guidelines for designing efficient BSFs for preventing microwave leakage and may have applications in other filters or array antennas using periodicity. Full article
(This article belongs to the Section Microwave and Wireless Communications)
►▼ Show Figures

Figure 1

36 pages, 2307 KB  
Article
From Energy Efficiency to Energy Intelligence: Power Electronics as the Cognitive Layer of the Energy Transition
by Nikolay Hinov
Electronics 2025, 14(23), 4673; https://doi.org/10.3390/electronics14234673 - 27 Nov 2025
Cited by 3 | Viewed by 1404
Abstract
The exponential growth of artificial intelligence (AI), electrified transport, and renewable generation is accelerating a structural shift in how societies produce, deliver, and consume electricity. We argue that the next frontier is not incremental efficiency but Energy Intelligence (EI): the embedding of predictive [...] Read more.
The exponential growth of artificial intelligence (AI), electrified transport, and renewable generation is accelerating a structural shift in how societies produce, deliver, and consume electricity. We argue that the next frontier is not incremental efficiency but Energy Intelligence (EI): the embedding of predictive analytics, adaptive control, and material-aware design directly into power-conversion hardware. In this view, power electronics functions as the cognitive layer that links digital intelligence to the physical flow of energy. Wide-bandgap (WBG) semiconductors—gallium nitride (GaN) and silicon carbide (SiC)—provide the material foundation for higher switching frequencies, superior power density, and real-time controllability, enabling compact and efficient converters for data-centers, EV charging, and grid-interactive resources. We formalize an EI reference architecture (predictive, adaptive, material-efficient, data-driven), review the convergence of AI methods with converter design and operation, and outline a GaN/SiC-enabled data-center power path as an illustrative case. Finally, we examine sustainability and sovereignty, highlighting exposure to critical materials (Ga, Si, In, rare earths) and proposing a roadmap that integrates technology, policy, and education. By reframing power electronics as an intelligent, learning infrastructure, this work sets an agenda for systems that are not only efficient but also self-optimizing, explainable, and resilient. Full article
(This article belongs to the Special Issue Section Collection Series: New Horizons and Recent Advances in Power Electronics)
►▼ Show Figures

Figure 1

19 pages, 3438 KB  
Article
Geometry-Aware Cross-Modal Translation with Temporal Consistency for Robust Multi-Sensor Fusion in Autonomous Driving
by Zhengyi Lu, Jinxiang Pang and Zhehai Zhou
Electronics 2025, 14(23), 4663; https://doi.org/10.3390/electronics14234663 - 27 Nov 2025
Cited by 1 | Viewed by 1462
Abstract
Intelligent Transportation Systems (ITSs), particularly autonomous driving, face critical challenges when sensor modalities fail due to adverse conditions or hardware malfunctions, causing severe perception degradation that threatens system-wide reliability. We present a unified geometry-aware cross-modal translation framework that synthesizes missing sensor data while [...] Read more.
Intelligent Transportation Systems (ITSs), particularly autonomous driving, face critical challenges when sensor modalities fail due to adverse conditions or hardware malfunctions, causing severe perception degradation that threatens system-wide reliability. We present a unified geometry-aware cross-modal translation framework that synthesizes missing sensor data while maintaining temporal consistency and quantifying uncertainty. Our pipeline enforces 92.7% frame-to-frame stability via an optical-flow-guided spatio-temporal module with smoothness regularization, preserves fine-grained 3D geometry through pyramid-level multi-scale alignment constrained by the Chamfer distance, surface normals, and edge consistency, and ultimately delivers dropout-tolerant perception by adaptively fusing multi-modal cues according to pixel-wise uncertainty estimates. Extensive evaluation on KITTI-360, nuScenes, and a newly collected Real-World Sensor Failure dataset demonstrates state-of-the-art performance: 35% reduction in Chamfer distance, 5% improvement in BEV (bird’s eye view) segmentation mIoU (mean Intersection over Union) (79.3%), and robust operation maintaining mIoU under complete sensor loss for 45+ s. The framework achieves real-time performance at 17 fps with 57% fewer parameters than competing methods, enabling deployment-ready sensor-agnostic perception for safety-critical autonomous driving applications. Full article
(This article belongs to the Special Issue Future Trends and Challenges of Ubiquitous Computing and Smart Systems, 2nd Edition)
►▼ Show Figures

Figure 1

24 pages, 9332 KB  
Article
Controlled Operation of Table ASSIST-EW Motion Assisting Device
by Earnest Ugonna Ofonaike and Marco Ceccarelli
Electronics 2025, 14(23), 4674; https://doi.org/10.3390/electronics14234674 - 27 Nov 2025
Viewed by 565
Abstract
Table ASSIST-EW is a lightweight, portable, and ergonomic exoskeletal device that is designed to support upper limb rehabilitation and to facilitate regular exercise in elderly users. Targeting the elbow and wrist joints, the device delivers smooth controlled assistance through a cable-driven actuation system [...] Read more.
Table ASSIST-EW is a lightweight, portable, and ergonomic exoskeletal device that is designed to support upper limb rehabilitation and to facilitate regular exercise in elderly users. Targeting the elbow and wrist joints, the device delivers smooth controlled assistance through a cable-driven actuation system that mimics natural muscle–tendon action. The system works with a scalable modular control architecture that enables the regulation of joint motion across a range of user needs and therapeutic contexts. The control design integrates force and motion feedback to implement assist-as-needed strategies, ensuring both safety and adaptability. Built on a bioinspired mechanical framework with revolute joint alignment and a soft inner interface for enhanced comfort, the device accommodates varied arm geometries and motion patterns. Simulation of key parameters—torque, stress, and energy demands—informed component selection and controller tuning. Experimental validation results confirm consistent performance across passive, active–assistive, and resistive control modes. Full article
(This article belongs to the Special Issue Advances in Kinematic Planning and Dynamic Control of Intelligent Robots)
►▼ Show Figures

Figure 1

41 pages, 3435 KB  
Systematic Review
Exploring the Educational Applications of Large Language Models: A Systematic Review and Topic Analysis
by Bianca-Raluca Cibu, Liliana Crăciun, Anca Gabriela Molănescu and Liviu-Adrian Cotfas
Electronics 2025, 14(23), 4683; https://doi.org/10.3390/electronics14234683 - 27 Nov 2025
Cited by 8 | Viewed by 3945
Abstract
In light of the accelerated growth of artificial intelligence (AI), large language models (LLMs) have become a central topic of interest in scientific research and practical applications across various fields. The present paper aims to perform a comprehensive systematic review of the scientific [...] Read more.
In light of the accelerated growth of artificial intelligence (AI), large language models (LLMs) have become a central topic of interest in scientific research and practical applications across various fields. The present paper aims to perform a comprehensive systematic review of the scientific literature on LLMs in education published between 2023 and 2024, based on a dataset from the Web of Science, which includes 507 documents from 322 sources. The accelerated dynamics of research in this field are confirmed by the high annual growth rate of 369.66%. The study identifies the themes presented in the scientific literature by using thematic maps and analyzing the evolution of said thematic maps. In addition, Latent Dirichlet Allocation (LDA) and BERTopic are used to outline the research field more clearly. Due to LDA’s ability to discover high-level research topics using probabilistic discovery and BERTopic’s ability to capture deeper semantic patterns and the emergence of various topics by searching, this paper first identifies the main research topics in the extracted dataset, which are then discussed in the paper through a review of applications. As a result, a range of applications are discovered in areas related to teaching and learning, academic assessment, integrity, academic feedback, medical education, ethics, bias, regulation, and social challenges. The conclusions provide a roadmap for researchers, practitioners and stakeholders in highlighting the current situation of LLMs in educational practice, while opening the door for future explorations in this domain. Full article
(This article belongs to the Section Artificial Intelligence)
►▼ Show Figures

Figure 1

19 pages, 2140 KB  
Article
AI-Driven Adaptive Segmentation of Timed Up and Go Test Phases Using a Smartphone
by Muntazir Rashid, Arshad Sher, Federico Villagra Povina and Otar Akanyeti
Electronics 2025, 14(23), 4650; https://doi.org/10.3390/electronics14234650 - 26 Nov 2025
Viewed by 1301
Abstract
The Timed Up and Go (TUG) test is a widely used clinical tool for assessing mobility and fall risk in older adults and individuals with neurological or musculoskeletal conditions. While it provides a quick measure of functional independence, traditional stopwatch-based timing offers only [...] Read more.
The Timed Up and Go (TUG) test is a widely used clinical tool for assessing mobility and fall risk in older adults and individuals with neurological or musculoskeletal conditions. While it provides a quick measure of functional independence, traditional stopwatch-based timing offers only a single completion time and fails to reveal which movement phases contribute to impairment. This study presents a smartphone-based system that automatically segments the TUG test into distinct phases, delivering objective and low-cost biomarkers of lower-limb performance. This approach enables clinicians to identify phase-specific impairments in populations such as individuals with Parkinson’s disease, and older adults, supporting precise diagnosis, personalized rehabilitation, and continuous monitoring of mobility decline and neuroplastic recovery. Our method combines adaptive preprocessing of accelerometer and gyroscope signals with supervised learning models (Random Forest, Support Vector Machine (SVM), and XGBoost) using statistical features to achieve continuous phase detection and maintain robustness against slow or irregular gait, accommodating individual variability. A threshold-based turn detection strategy captures both sharp and gradual rotations. Validation against video ground truth using group K-fold cross-validation demonstrated strong and consistent performance: start and end points were detected in 100% of trials. The mean absolute error for total time was 0.42 s (95% CI: 0.36–0.48 s). The average error across phases (stand, walk, turn) was less than 0.35 s, and macro F1 scores exceeded 0.85 for all models, with the SVM achieving the highest score of 0.882. Combining accelerometer and gyroscope features improved macro F1 by up to 12%. Statistical tests (McNemar, Bowker) confirmed significant differences between models, and calibration metrics indicated reliable probabilistic outputs (ROC-AUC > 0.96, Brier score < 0.08). These findings show that a single smartphone can deliver accurate, interpretable, and phase-aware TUG analysis without complex multi-sensor setups, enabling practical and scalable mobility assessment for clinical use. Full article
(This article belongs to the Special Issue Emerging Trend in Intelligent Activity Recognition and Gait Monitoring in Real Environments)
►▼ Show Figures

Figure 1

11 pages, 428 KB  
Article
RMF-A: An Availability Assurance Framework for Quantitative Evaluation of Operational Resilience
by Cheon-Ho Min and Jin Kwak
Electronics 2025, 14(23), 4644; https://doi.org/10.3390/electronics14234644 - 26 Nov 2025
Viewed by 954
Abstract
Recent data center incidents have revealed that certification under ISO 22301 and ISO/IEC 27001 does not guarantee real operational resilience. This study presents the Availability Assurance Framework (RMF-A), an extension of the NIST Risk Management Framework that introduces an Availability Assurance Phase. RMF-A [...] Read more.
Recent data center incidents have revealed that certification under ISO 22301 and ISO/IEC 27001 does not guarantee real operational resilience. This study presents the Availability Assurance Framework (RMF-A), an extension of the NIST Risk Management Framework that introduces an Availability Assurance Phase. RMF-A combines ISO-based management controls with NIST’s evidence-driven assessment using the Availability Evidence Model (AEM) and the Availability Assurance Index (AAI). AEM defines measurable indicators—recovery rate (RR), recovery time (RTO), and Detection Effectiveness (DET)—and AAI aggregates them into a quantitative assurance score. Validation using three open datasets—Google Cluster Trace, Azure Cloud Trace, and LANL HPC Logs—showed consistent assurance results: Google (AAI = 0.758, ATO-Conditional), Azure (AAI = 0.720, ATO-Conditional), and LANL HPC (AAI = 0.744, ATO-Conditional). The results confirm that RMF-A provides a reproducible, evidence-based approach to quantify operational resilience and ensure continuous availability. Full article
(This article belongs to the Special Issue AI-Driven Cybersecurity, Resilience, and Trust Frameworks for Future Urban IoT Systems)
►▼ Show Figures

Figure 1

21 pages, 4092 KB  
Article
Enabling Scalable and Manufacturable Large-Scale Antenna Arrays Through Hexagonal Subarray Implementation over Goldberg Polyhedra
by Santiago Loza-Morcillo and José Luis Blanco-Murillo
Electronics 2025, 14(23), 4618; https://doi.org/10.3390/electronics14234618 - 25 Nov 2025
Cited by 1 | Viewed by 1357
Abstract
We introduce a scalable and manufacturable approach to conformal large-scale antenna arrays, leveraging Goldberg Polyhedra configurations with hexagonal subarrays to enable cost-effective, high-performance beam steering. Planar array designs face challenges in phase control and beam deformation when steering away from the broadside, leading [...] Read more.
We introduce a scalable and manufacturable approach to conformal large-scale antenna arrays, leveraging Goldberg Polyhedra configurations with hexagonal subarrays to enable cost-effective, high-performance beam steering. Planar array designs face challenges in phase control and beam deformation when steering away from the broadside, leading to increased beamwidth and degraded angular resolution. Our near-spherical Goldberg structures offer a fabrication-friendly, periodic architecture that supports industrial scalability while enabling efficient 360° digital beamforming with minimal distortion. Simulation results confirm significant reductions in sidelobe levels and improved energy concentration, providing enhanced multibeam capabilities and simplified digital beamforming (DBF) control. This approach paves the way for next-generation radar and satellite systems requiring precise directional control, minimal interference, and robust, flexible beam steering performance. Full article
(This article belongs to the Special Issue Recent Advancements in Antenna Technologies: Novel Design Techniques and Emerging Applications)
►▼ Show Figures

Figure 1

23 pages, 8756 KB  
Article
Application and Development of a Double Asymmetric Voltage Modulation on a Resonant Dual Active Bridge
by Mattia Vogni, Juan L. Bellido, Fausto Stella, Leonardo Stefanini, Claudio Bianchini and Vicente Esteve
Electronics 2025, 14(23), 4625; https://doi.org/10.3390/electronics14234625 - 25 Nov 2025
Cited by 4 | Viewed by 889
Abstract
The growing market penetration of Electric Vehicles (EVs) requires very efficient bidirectional on-board chargers. These converters must allow the power transfer from the grid to the battery of the vehicle and vice versa, since Vehicle to Grid (V2G) applications enable a mitigation of [...] Read more.
The growing market penetration of Electric Vehicles (EVs) requires very efficient bidirectional on-board chargers. These converters must allow the power transfer from the grid to the battery of the vehicle and vice versa, since Vehicle to Grid (V2G) applications enable a mitigation of the peak demand and help regulate both the voltage and the frequency of the grid. In this paper, an innovative double asymmetric modulation was studied and applied to a resonant Dual Active Bridge (DAB), CLLC resonant filter configuration. The results of the study showed a significant efficiency boost and an easier controllability of the converter with respect to more traditional modulations or variable frequency techniques, maintaining Zero-Voltage Switching (ZVS) conditions for all the switches in a wide operating range, from 28 to 100% of the maximum power (4–14 kW). A map of optimum points, where converter losses are minimized, is calculated offline through an algorithm in MATLAB R2024a and a proper interpolation between these points allows any output power for each possible voltage level of the battery to be achieved: from 250 V up to 400 V. The modulations are compared and evaluated through simulations carried out in PLECS, both offline and using hardware-in-the-loop (HIL), as well as through experimental tests. Full article
(This article belongs to the Special Issue Advances in Power Converter Design, Control and Applications, 2nd Edition)
►▼ Show Figures

Figure 1

20 pages, 4144 KB  
Article
Model Predictive Control Using the Improved Hovorka Model for the Regulation of Blood Glucose Levels in Type 1 Diabetes
by Iqra Shafeeq Mughal, Stefan Koch, Luca Patanè, Martin Steinberger, Riccardo Caponetto and Nebojša Koledin
Electronics 2025, 14(23), 4585; https://doi.org/10.3390/electronics14234585 - 23 Nov 2025
Cited by 1 | Viewed by 1637
Abstract
Type 1 diabetes is an autoimmune disease that occurs when the immune system unintentionally attacks and damages β cells in the pancreas, reducing the organ’s ability to produce insulin. An artificial pancreas is a technology that uses a pump to inject the appropriate [...] Read more.
Type 1 diabetes is an autoimmune disease that occurs when the immune system unintentionally attacks and damages β cells in the pancreas, reducing the organ’s ability to produce insulin. An artificial pancreas is a technology that uses a pump to inject the appropriate amount of insulin subcutaneously after analysing information collected by sensors, including continuous blood glucose monitoring. Over the past thirty years, several methods for controlling an artificial pancreas have been investigated in clinical and simulation environments. The improved Hovorka model, a comprehensive nonlinear model that explains the effects of insulin on transport, disposal and endogenous synthesis in both accessible and inaccessible compartments for blood glucose control by insulin administration, is used for this research. The presented model has the characteristics of a switching nonlinear system. The work proposes to analyse different nonlinear control strategies for blood glucose regulation and shows the effectiveness of the linear model predictive control strategy compared to other nonlinear controllers used in the literature. Full article
(This article belongs to the Section Systems & Control Engineering)
►▼ Show Figures

Figure 1

38 pages, 25113 KB  
Article
A Two-Stage End-to-End Framework for Robust Scene Text Spotting with Self-Calibrated Detection and Contextual Recognition
by Yuning Cheng, Jinhong Huang, Io San Tai, Subrota Kumar Mondal, Tianqi Wang and Hussain Mohammed Dipu Kabir
Electronics 2025, 14(23), 4594; https://doi.org/10.3390/electronics14234594 - 23 Nov 2025
Cited by 2 | Viewed by 1959
Abstract
End-to-end scene text detection and recognition, which involves detecting and recognizing text in natural images, still faces significant challenges, particularly in handling text of arbitrary shapes, complex backgrounds, and computational efficiency requirements. This paper proposes a novel and viable end-to-end OCR framework that [...] Read more.
End-to-end scene text detection and recognition, which involves detecting and recognizing text in natural images, still faces significant challenges, particularly in handling text of arbitrary shapes, complex backgrounds, and computational efficiency requirements. This paper proposes a novel and viable end-to-end OCR framework that synergistically combines a powerful detection network with advanced recognition models. For text detection, we develop a method called Text Contrast Self-Calibrated Network (TextCSCN), which employs pixel-wise supervised contrastive learning to extract more discriminative features. TextCSCN addresses long-range dependency modeling and limited receptive field issues through self-calibrated convolutions and Global Convolutional Networks (GCNs). We further introduce an efficient Mamba-based bidirectional module for boundary refinement, enhancing both accuracy and speed. For text recognition, our framework employs a Swin Transformer backbone with Bidirectional Feature Pyramid Networks (BiFPNs) for optimized multi-scale feature extraction. We propose a Pre-Gated Contextual Attention Gate (PCAG) mechanism to effectively fuse visual and linguistic information while minimizing noise and uncertainty in multi-modal integration. Experiments on challenging benchmarks including TotalText and CTW1500 demonstrate the effectiveness of our approach. Our detection module achieves state-of-the-art performance with an F-score of 88.21% on TotalText, and the complete end-to-end system shows comparable improvements in recognition accuracy, establishing new benchmarks for scene text spotting. Full article
(This article belongs to the Special Issue Exploring Edge AI: Architectures, Algorithms, and the Role of Edge–Cloud Cooperation for Scalable AI Systems)
►▼ Show Figures

Figure 1

13 pages, 835 KB  
Article
Layer-Pipelined CNN Accelerator Design on 2.5D FPGAs
by Mengxuan Wang and Chang Wu
Electronics 2025, 14(23), 4587; https://doi.org/10.3390/electronics14234587 - 23 Nov 2025
Cited by 1 | Viewed by 1194
Abstract
With the rapid advancement of 2.5D FPGA technology, the integration of multiple FPGA dies enables larger design capacity and higher computing power. This progress provides a high-speed hardware platform well-suited for neural network acceleration. In this paper, we present a high-performance accelerator design [...] Read more.
With the rapid advancement of 2.5D FPGA technology, the integration of multiple FPGA dies enables larger design capacity and higher computing power. This progress provides a high-speed hardware platform well-suited for neural network acceleration. In this paper, we present a high-performance accelerator design for large-scale neural networks on 2.5D FPGAs. First, we propose a layer pipeline architecture that utilizes multiple accelerator cores, each equipped with individual high-bandwidth DDR memory. To address inter-die data dependencies, we introduce a block convolution mechanism that enables independent and efficient computation across dies. Furthermore, we propose a design space exploration scheme to optimize computational efficiency under resource constraints. Experimental results demonstrate that our proposed accelerator achieves 4860.87 GOPS when running VGG-16 on the Alveo U250 board, significantly outperforming existing layer pipeline designs on the same platform. Full article
(This article belongs to the Special Issue Advances in High-Performance and Parallel Computing)
►▼ Show Figures

Figure 1

31 pages, 1411 KB  
Article
A Source-to-Source Compiler to Enable Hybrid Scheduling for High-Level Synthesis
by Yuhan She, Yanlong Huang, Jierui Liu, Ray C. C. Cheung and Hong Yan
Electronics 2025, 14(23), 4578; https://doi.org/10.3390/electronics14234578 - 22 Nov 2025
Viewed by 1015
Abstract
High-Level Synthesis (HLS) has gained considerable attention for its ability to quickly generate hardware descriptions from untimed specifications. Most state-of-the-art commercial HLS tools employ static scheduling, which excels in compute-intensive applications but struggles with control-dominant designs. While some open-source tools propose dynamic and [...] Read more.
High-Level Synthesis (HLS) has gained considerable attention for its ability to quickly generate hardware descriptions from untimed specifications. Most state-of-the-art commercial HLS tools employ static scheduling, which excels in compute-intensive applications but struggles with control-dominant designs. While some open-source tools propose dynamic and hybrid scheduling techniques to synthesize dataflow-like architectures to improve speed, they lack well-established optimizations from static scheduling like datapath optimization and resource sharing, leading to frequency degradation and area overhead. Moreover, existing hybrid scheduling relies on extra dynamic synthesis support, either by dynamic or static HLS tools, and thereby loses generality. In this work, we propose another solution to achieve hybrid scheduling: a source-to-source compiler that exposes dynamism at the source code level, which reduces both frequency and area overhead while remaining fully compatible with modern static HLS tools without needing extra dynamic synthesis support. Experiments show significant improvements (1.26× speedup) on wall clock time (WCT) compared to VitisHLS and a better area–frequency–latency trade-off compared to dynamic (1.83× WCT speedup and 0.46× area) and hybrid (2.14× WCT speedup and 0.72× area) scheduling-based tools. Full article
(This article belongs to the Special Issue Emerging Applications of FPGAs and Reconfigurable Computing System)
►▼ Show Figures

Figure 1

33 pages, 1391 KB  
Review
Hyperspectral Imaging System Applications in Healthcare
by Krzysztof Wołk and Agnieszka Wołk
Electronics 2025, 14(23), 4575; https://doi.org/10.3390/electronics14234575 - 22 Nov 2025
Cited by 9 | Viewed by 3875
Abstract
Hyperspectral imaging (HSI) is a swiftly developing intraoperative and diagnostic technique in several clinical specialties. By monitoring oxygenation and biochemical markers, it helps with tissue viability, burn depth measurement, wound healing, and tumor detection. HSI facilitates real-time, harmless diagnosis throughout surgeries or outpatient [...] Read more.
Hyperspectral imaging (HSI) is a swiftly developing intraoperative and diagnostic technique in several clinical specialties. By monitoring oxygenation and biochemical markers, it helps with tissue viability, burn depth measurement, wound healing, and tumor detection. HSI facilitates real-time, harmless diagnosis throughout surgeries or outpatient settings, and allows for the detection of tumor boundaries with over 90% accuracy, according to clinical studies. Originally developed for remote sensing and aerospace applications, HSI has rapidly evolved and found increasing relevance across diverse sectors, including agriculture, environmental monitoring, food safety, pharmaceuticals, defense, and especially medical diagnostics. This review explores the origins, development, and expanding applications of HSI, with a particular emphasis on its role in healthcare. It discusses the operational principles and unique features of hyperspectral systems, such as their ability to produce spectral data cubes, perform non-destructive analysis, and integrate with emerging technologies like artificial intelligence and drone-based platforms. By comparing hyperspectral imaging to traditional and multispectral techniques, the review highlights its superior spectral resolution and versatility. Key challenges, including data volume, sensor calibration, and real-time processing, are also addressed. Finally, emerging trends such as miniaturization, integration with the Internet of Things, and sustainable system designs are examined, offering insights into the future directions and interdisciplinary potentials of HSI in both scientific research and practical applications. Full article
(This article belongs to the Special Issue Hyperspectral Imaging: Technologies and Applications)
►▼ Show Figures

Figure 1

36 pages, 3549 KB  
Article
Feasibility of Large-Scale Electric Vehicle Deployment in Islanded Grids: The Canary Islands Case
by Alejandro García García, Víctor Rubio Matilla, Juan Diego López Arquillo and Cristiana Oliveira
Electronics 2025, 14(23), 4579; https://doi.org/10.3390/electronics14234579 - 22 Nov 2025
Cited by 1 | Viewed by 1626
Abstract
The present integration of electric vehicles into everyday life has the potential to redefine current standards of urban mobility. However, the territorial impact of this deployment demands a multiscale effort to ensure both efficient and sustainable performance; this is even more necessary in [...] Read more.
The present integration of electric vehicles into everyday life has the potential to redefine current standards of urban mobility. However, the territorial impact of this deployment demands a multiscale effort to ensure both efficient and sustainable performance; this is even more necessary in a disconnected system like an island. This article addresses the possibility of transforming the existing fossil-fuel-based infrastructure within Europe’s outermost regions into an electric vehicle charging network, with particular emphasis on the Canary Islands’ strategic plans. Using official datasets from Red Eléctrica de España (REE), IDAE, and the Canary Islands’ Energy Transition Plan (PTECan), we develop three scenarios (2025 baseline, 2030, and 2040) to quantify the additional electricity demand, peak load requirements, charging infrastructure needs, and associated greenhouse gas emissions. The methodology combines EV fleet projections, the driving patterns of residents and tourists, and vehicle efficiency data to estimate yearly electricity demand and hourly charging loads. The carbon intensity profiles of each island’s grid are used to calculate well-to-wheel emissions of EVs, benchmarked against internal combustion engine vehicles. The results indicate that achieving 250,000 EVs by 2030 would increase electricity demand by 1.1–1.4 TWh/year (+8–12% of current consumption), requiring approximately 25,000–30,000 public charging points. EV emissions range from 90 to 150 gCO2/km depending on charging time, compared to 160–190 gCO2/km for ICE vehicles. Smart charging and vehicle-to-grid integration could mitigate 15–25% of peak load increases, reducing the curtailment of renewables and deferring grid investments. A comparative analysis with Zealand highlights policy synergies and differences in insular versus continental grids. The findings confirm that large-scale EV adoption in the Canary Islands is technically feasible, but quite difficult, as it requires the deep, coordinated planning of renewable expansion, storage, and a charging infrastructure. BEV WTW advantages become unequivocal once the average grid carbon intensity falls below ≈0.8–0.9 tCO2/MWh, underscoring the primacy of accelerated renewable build-out and demand-side flexibility. Despite uncertainties in adoption and technology trajectories, the approach is transparent and reproducible with official datasets, providing a transferable planning tool for other islanded systems and mainland Europe. The proposed method demonstrates its usefulness in direct linking electrification scenarios with the real capacity of the electricity system, allowing the identification of very critical integration thresholds and guiding evidence-based planning decisions. Full article
(This article belongs to the Special Issue Advances in Electric Vehicle Technology)
►▼ Show Figures

Figure 1

27 pages, 659 KB  
Review
From Vulnerability to Robustness: A Survey of Patch Attacks and Defenses in Computer Vision
by Xinyun Liu and Ronghua Xu
Electronics 2025, 14(23), 4553; https://doi.org/10.3390/electronics14234553 - 21 Nov 2025
Cited by 1 | Viewed by 2213
Abstract
Adversarial patch attacks have emerged as a powerful and practical threat to machine learning models in vision-based tasks. Unlike traditional perturbation-based adversarial attacks, which often require imperceptible changes to the entire input, patch attacks introduce localized and visible modifications that can consistently mislead [...] Read more.
Adversarial patch attacks have emerged as a powerful and practical threat to machine learning models in vision-based tasks. Unlike traditional perturbation-based adversarial attacks, which often require imperceptible changes to the entire input, patch attacks introduce localized and visible modifications that can consistently mislead deep neural networks across varying conditions. Their physical realizability makes them particularly concerning for real-world security-critical applications. In response, a growing body of research has proposed diverse defense strategies, including input preprocessing, robust model training, detection-based approaches, and certified defense mechanisms. In this paper, we provide a comprehensive review of patch-based adversarial attacks and corresponding defense techniques. First, we introduce a new task-oriented taxonomy that systematically categorizes patch attack methods according to their downstream vision applications (e.g., classification, detection, segmentation), and then we summarize defense mechanisms based on three major strategies: Patch Localization and Removal-based Defenses, Input Transformation and Reconstruction-based Defenses, Model Modification and Training-based Defenses. This unified framework provides an integrated perspective that bridges attack and defense research. Furthermore, we highlight open challenges, such as balancing robustness and model utility, addressing adaptive attackers, and ensuring physical-world resilience. Finally, we outline promising research directions to inspire future work toward building trustworthy and robust vision systems against patch-based adversarial threats. Full article
(This article belongs to the Special Issue Artificial Intelligence Safety and Security)
►▼ Show Figures

Figure 1

22 pages, 5988 KB  
Article
Research on Battery Aging and User Revenue of Electric Vehicles in Vehicle-to-Grid (V2G) Scenarios
by Zhiyu Zhao, Shuaihao Kong, Bo Bo, Xuemei Li, Ling Hao, Fei Xu and Lei Chen
Electronics 2025, 14(23), 4567; https://doi.org/10.3390/electronics14234567 - 21 Nov 2025
Cited by 3 | Viewed by 2277
Abstract
With the development of vehicle-to-grid (V2G) technology, electric vehicles (EVs) are increasingly participating in grid interactions. However, V2G-induced energy consumption and battery aging intensify range anxiety among users, reduce participation willingness, and decrease discharge capacity and revenue due to capacity loss. In this [...] Read more.
With the development of vehicle-to-grid (V2G) technology, electric vehicles (EVs) are increasingly participating in grid interactions. However, V2G-induced energy consumption and battery aging intensify range anxiety among users, reduce participation willingness, and decrease discharge capacity and revenue due to capacity loss. In this study, aging models for power batteries in electric passenger vehicles and electric trucks are established. A time-of-use electricity price model and an economic model considering battery aging costs are constructed. Two scenarios were established for daily use and V2G operation. The impacts of different scenarios and charging/discharging patterns on battery life and user profit are analyzed. The results indicate that the additional V2G discharging process increases the cyclic aging rate of EV batteries. Within the studied parameter ranges, the cyclic aging rate increased by 5.89% for electric passenger vehicles and 3.72% for electric trucks, respectively. Additionally, the initial V2G revenue may struggle to cover early-stage battery aging costs, but the subsequent slowdown in degradation may eventually offset these costs. With appropriate charging and discharging strategies, the maximum revenue per year reaches 18,200 CNY for electric trucks and 5600 CNY for electric passenger vehicles. This study may provide theoretical support for optimizing EV charging/discharging strategies and formulating policies in V2G scenarios. Full article
(This article belongs to the Special Issue Electro-Thermal Modelling, Status Estimation and Thermal Management of Electric Vehicles, 2nd Edition)
►▼ Show Figures

Figure 1

26 pages, 952 KB  
Article
From Forecasting to Foresight: Building an Autonomous O&M Brain for the New Power System Based on a Cognitive Digital Twin
by Xufeng Wu, Zuowei Chen, Hefang Jiang, Shoukang Luo, Yi Zhao, Dongwei Zhao, Peiyao Dang, Jiajun Gao, Lin Lin and Hao Wang
Electronics 2025, 14(22), 4537; https://doi.org/10.3390/electronics14224537 - 20 Nov 2025
Cited by 8 | Viewed by 1601
Abstract
Despite notable advances in load forecasting and fault detection, current power system operation and maintenance (O&M) technologies remain fragmented into independent and primarily reactive modules. Load forecasting estimates future demand, whereas fault detection identifies whether abnormal conditions exist in the present state. This [...] Read more.
Despite notable advances in load forecasting and fault detection, current power system operation and maintenance (O&M) technologies remain fragmented into independent and primarily reactive modules. Load forecasting estimates future demand, whereas fault detection identifies whether abnormal conditions exist in the present state. This paper proposes a unified and proactive Cognitive Digital Twin (CDT) system. Unlike traditional data-driven approaches, the CDT integrates Large Language Models (LLMs) and Knowledge Graphs (KGs) as cognitive cores to enable deeper reasoning and context-aware decision-making. The CDT system not only mirrors the physical grid but also acts as an intelligent O&M engine capable of understanding, reasoning, predicting, and self-diagnosing. The core innovation lies in prediction-based anomaly detection. The system first estimates the expected healthy state of the grid at future time steps and then compares real-time monitoring data against these predictions to identify incipient anomalies. This enables genuine foresight rather than simple reactive detection. By orchestrating advanced analytical modules, including CNN–LSTM hybrid models and optimization algorithms, the CDT supports autonomous O&M operations with transparent and explainable decision-making. These capabilities enhance grid resilience and improve the system’s capacity for self-healing. Full article
(This article belongs to the Special Issue Data Analysis and Data Fusion in System Identification and Measurements)
►▼ Show Figures

Figure 1

24 pages, 9193 KB  
Article
Leveraging Software-Defined Networking for Secure and Resilient Real-Time Power Sharing in Multi-Microgrid Systems
by Rawan A. Taha, Ahmed Aghmadi, Sara H. Moustafa and Osama A. Mohammed
Electronics 2025, 14(22), 4518; https://doi.org/10.3390/electronics14224518 - 19 Nov 2025
Cited by 2 | Viewed by 1051
Abstract
Cyber-physical power systems integrate sensing, communication, and control, ensuring power system resiliency and security, particularly in clustered networked microgrids. Software-Defined Networking (SDN) provides a suitable foundation by centralizing policy, enforcing traffic isolation, and adopting a deny-by-default policy in which only explicitly authorized flows [...] Read more.
Cyber-physical power systems integrate sensing, communication, and control, ensuring power system resiliency and security, particularly in clustered networked microgrids. Software-Defined Networking (SDN) provides a suitable foundation by centralizing policy, enforcing traffic isolation, and adopting a deny-by-default policy in which only explicitly authorized flows are admitted. This paper proposes and experimentally validates a cyber-physical architecture that couples three DC microgrids through an SDN backbone to deliver rapid, reliable, and secure power sharing under highly dynamic conditions, including pulsed-load disturbances. The cyber layer comprises four SDN switches that establish dedicated paths for protection messages, supervisory control commands, and high-rate sensor data streams. An OpenFlow controller administers flow-rule priorities, link monitoring, and automatic failover to preserve control command paths during disturbances and communication faults. Resiliency is further assessed by subjecting the network to a deliberate denial-of-service (DoS) attack, where deny-by-default policies prevent unauthorized traffic while maintaining essential control flows. Performance is quantified through packet captures, which include end-to-end delay, jitter, and packet loss percentage, alongside synchronized electrical measurements from high-resolution instrumentation. Results show that SDN-enforced paths, combined with coordinated multi-microgrid control, maintain accurate power sharing. A validated, hardware testbed demonstration substantiates a scalable, co-designed communication-and-control framework for next-generation cyber-physical DC multi-microgrid deployments. Full article
(This article belongs to the Special Issue Efficient and Resilient DC Energy Distribution Systems)
►▼ Show Figures

Figure 1

22 pages, 83077 KB  
Article
Comparative Analysis of SiC-Based Isolated Bidirectional DC/DC Converters for a Modularized Off-Board EV Charging System with a Bipolar DC Link
by Kaushik Naresh Kumar, Rafał Miśkiewicz, Przemysław Trochimiuk, Jacek Rąbkowski and Dimosthenis Peftitsis
Electronics 2025, 14(22), 4522; https://doi.org/10.3390/electronics14224522 - 19 Nov 2025
Cited by 4 | Viewed by 1862
Abstract
The choice of a suitable isolated and bidirectional DC/DC converter (IBDC) topology is an important step in the design of a bidirectional electric vehicle (EV) charging system. In this context, six 10 kW rated silicon carbide (SiC) metal–oxide–semiconductor field-effect transistor (MOSFET)-based dual-active bridge [...] Read more.
The choice of a suitable isolated and bidirectional DC/DC converter (IBDC) topology is an important step in the design of a bidirectional electric vehicle (EV) charging system. In this context, six 10 kW rated silicon carbide (SiC) metal–oxide–semiconductor field-effect transistor (MOSFET)-based dual-active bridge (DAB) converter topologies, supplied by a +750/0/−750 V bipolar DC link, are analyzed and compared in this article. The evaluation criteria include the required volt-ampere semiconductor ratings, loss distribution, efficiency, and thermal considerations of the considered converter configurations. The IBDC topologies are compared based on the observations and results obtained from theoretical analysis, electro-thermal simulations, and experiments, considering the same voltage and power conditions. The advantages and disadvantages of the topologies, in terms of the considered evaluation criteria, are discussed. It is shown that the series-resonant (SR) input-series output-parallel (ISOP) full-bridge (FB) DAB converter configuration is the most suitable design choice for the considered EV charging application based on the chosen operating conditions and evaluation criteria. Full article
(This article belongs to the Special Issue DC–DC Power Converter Technologies for Energy Storage Integration)
►▼ Show Figures

Figure 1

24 pages, 4507 KB  
Article
Ultra-Short-Term Power Prediction for Distributed Photovoltaics Based on Time-Series LLMs
by Chen Lv, Hang Fan, Zuhan Zhang, Menghua Fan, Wencai Run, Liuqing Yang, Yuying Yang and Dunnan Liu
Electronics 2025, 14(22), 4519; https://doi.org/10.3390/electronics14224519 - 19 Nov 2025
Cited by 6 | Viewed by 1405
Abstract
Distributed photovoltaic power generation is volatile and intermittent, and its power generation is usually difficult to accurately predict. Previous studies have focused on physical or mathematical modeling methods, and it is difficult to grasp the complexity and variability of historical data, and the [...] Read more.
Distributed photovoltaic power generation is volatile and intermittent, and its power generation is usually difficult to accurately predict. Previous studies have focused on physical or mathematical modeling methods, and it is difficult to grasp the complexity and variability of historical data, and the prediction accuracy is limited. To address these challenges, this paper proposes Solar-LLM, a novel prediction framework that adapts a pre-trained Large Language Model (LLM) for time-series forecasting. By freezing the core LLM and reprogramming only its input and output layers, Solar-LLM efficiently translates numerical time-series data into a format the model can understand. This approach leverages the LLM’s inherent ability to capture long-term dependencies and complex patterns, enabling effective learning even from limited data. Experiments conducted on a dataset from five photovoltaic power stations show that Solar-LLM significantly outperforms baseline models, proving it to be a highly effective and feasible solution for distributed PV power prediction. Full article
►▼ Show Figures

Figure 1

21 pages, 3728 KB  
Article
A Multi-Core Benchmark Framework for Linux-Based Embedded Systems Using Synthetic Task-Set Generation
by Yixiao Xing, Yixiao Li and Hiroaki Takada
Electronics 2025, 14(22), 4515; https://doi.org/10.3390/electronics14224515 - 19 Nov 2025
Viewed by 1595
Abstract
Accurately evaluating multi-core embedded systems remains a major challenge, as existing benchmarking methods and tools fail to reproduce realistic workloads with inter-core contentions. This study introduces a benchmark framework for Linux-based embedded systems that integrates a synthetic task-set generation model capable of reproducing [...] Read more.
Accurately evaluating multi-core embedded systems remains a major challenge, as existing benchmarking methods and tools fail to reproduce realistic workloads with inter-core contentions. This study introduces a benchmark framework for Linux-based embedded systems that integrates a synthetic task-set generation model capable of reproducing both computational and contention characteristics observed in real-world applications. Applying this benchmark to three Linux kernel variants on a 16-core embedded platform, we have identified distinct scalability patterns and contention sensitivities among kernel configurations. The results mainly demonstrate the framework’s capability to reveal performance characteristics under Linux, but the proposed methodology itself has high portability and extendability by design to support various multi-core platforms including the RTOS-based ones. Full article
(This article belongs to the Special Issue Design and Implementation of Embedded Systems for Real-Time Applications)
►▼ Show Figures

Figure 1

29 pages, 6088 KB  
Article
Lightweight AI for Sensor Fault Monitoring
by Bektas Talayoglu, Jerome Vande Velde and Bruno da Silva
Electronics 2025, 14(22), 4532; https://doi.org/10.3390/electronics14224532 - 19 Nov 2025
Viewed by 3694
Abstract
Sensor faults can produce incorrect data and disrupt the operation of entire systems. In critical environments, such as healthcare, industrial automation, or autonomous platforms, these faults can lead to serious consequences if not detected early. This study explores how faults in MEMS microphones [...] Read more.
Sensor faults can produce incorrect data and disrupt the operation of entire systems. In critical environments, such as healthcare, industrial automation, or autonomous platforms, these faults can lead to serious consequences if not detected early. This study explores how faults in MEMS microphones can be classified using lightweight ML models suitable for devices with limited resources. A dataset was created for this work, including both real faults (normal, clipping, stuck, and spikes) caused by issues like acoustic overload and undervoltage, and synthetic faults (drift and bias). The goal was to simulate a range of fault behaviors, from clear malfunctions to more subtle signal changes. Convolutional Neural Networks (CNNs) and hybrid models that use CNNs for feature extraction with classifiers like Decision Trees, Random Forest, MLP, Extremely Randomized Trees, and XGBoost, were evaluated based on accuracy, F1-score, inference time, and model size towards real-time use in embedded systems. Experiments showed that using 2-s windows improved accuracy and F1-scores. These findings help design ML solutions for sensor fault classification in resource-limited embedded systems and IoT applications. Full article
(This article belongs to the Special Issue Advances of Artificial Intelligence and Vision Applications, 2nd Edition)
►▼ Show Figures

Figure 1

18 pages, 12569 KB  
Article
YOLO-DH: Robust Object Detection for Autonomous Vehicles in Adverse Weather
by Liye Jiang, Guanqun Ma, Weixuan Guo and Yunhao Sun
Electronics 2025, 14(22), 4476; https://doi.org/10.3390/electronics14224476 - 17 Nov 2025
Cited by 8 | Viewed by 1973
Abstract
Although autonomous driving technology has advanced rapidly in recent years, the intrinsic complexities of adverse weather conditions pose significant challenges to achieving high object detection accuracy, despite extensive research on perception systems in autonomous vehicles. Accordingly, this study proposes a YOLO-based network that [...] Read more.
Although autonomous driving technology has advanced rapidly in recent years, the intrinsic complexities of adverse weather conditions pose significant challenges to achieving high object detection accuracy, despite extensive research on perception systems in autonomous vehicles. Accordingly, this study proposes a YOLO-based network that integrates the DHNet residual network for image dehazing with wavelet-based channel attention modules, specifically tailored for target detection in autonomous driving under adverse weather conditions. The proposed network introduces a novel channel-value attention mechanism and sequentially applies wavelet-based soft-thresholding denoising techniques, thereby improving its sensitivity to discriminative channel information. We further introduce the first DHNet Dehazing Attention Module, which sequentially combines the DHNet residual network with the MixDehazeNet hybrid structural dehazing network. This integration synergistically improves the network’s image dehazing performance. To address dataset class imbalance, this study incorporates Adaptive Threshold Focal Loss (ATFL), which markedly improves training efficiency and model robustness. This optimization enhances the model’s generalization ability for target detection and classification tasks under challenging weather conditions. Experimental evaluations, including ablation studies and comparative analyses, demonstrate that the proposed method achieves substantial improvements in accuracy across three benchmark datasets. Full article
►▼ Show Figures

Figure 1

19 pages, 3485 KB  
Article
Predictive Risk-Aware Reinforcement Learning for Autonomous Vehicles Using Safety Potential
by Jinho Choi and Shiho Kim
Electronics 2025, 14(22), 4446; https://doi.org/10.3390/electronics14224446 - 14 Nov 2025
Cited by 4 | Viewed by 2399
Abstract
Safety remains a central challenge in autonomous driving: overly rigid safeguards can cause unnecessary stops and erode efficiency. Addressing this safety–efficiency trade-off requires specifying what behaviors to incentivize. In reinforcement learning, the reward provides that specification. Conventional reward surrogates—such as distance gaps and [...] Read more.
Safety remains a central challenge in autonomous driving: overly rigid safeguards can cause unnecessary stops and erode efficiency. Addressing this safety–efficiency trade-off requires specifying what behaviors to incentivize. In reinforcement learning, the reward provides that specification. Conventional reward surrogates—such as distance gaps and time-to-collision (TTC)—depend on instantaneous geometry and often miss unfolding multi-vehicle interactions, whereas sparse terminal rewards provide no intermediate guidance. Accordingly, we adapt Safety Potential (SP)—a short-horizon, time-weighted path-overlap forecast—into a dense reward-shaping term that provides a predictive risk-aware signal for anticipatory throttle/brake control. In the CARLA v0.9.14 roundabout environment, SP attains 94% success with 3% collisions; in percentage points, this is 16.00, 13.00, and 5.75 higher success and 18.75, 9.50, and 7.25 lower collisions than No-Safe, Distance, and TTC, respectively. Adding a lightweight reactive guard at inference further reduces collisions to below 1% without sacrificing success. These results indicate that injecting a predictive, overlap-based risk measure directly into the reward supplies temporally consistent safety cues and clarifies the trade-off between progress and risk in reinforcement-learning-based vehicle control. Full article
(This article belongs to the Special Issue Feature Papers in Electrical and Autonomous Vehicles, Volume 2)
►▼ Show Figures

Figure 1

14 pages, 2150 KB  
Article
A Flexible Multi-Core Hardware Architecture for Stereo-Based Depth Estimation CNNs
by Steven Colleman, Andrea Nardi-Dei, Marc C. W. Geilen, Sander Stuijk and Toon Goedemé
Electronics 2025, 14(22), 4425; https://doi.org/10.3390/electronics14224425 - 13 Nov 2025
Cited by 2 | Viewed by 708
Abstract
Stereo-based depth estimation is becoming more and more important in many applications like self-driving vehicles, earth observation, cartography, robotics and so on. Modern approaches to depth estimation employ artificial intelligence techniques, particularly convolutional neural networks (CNNs). However, stereo-based depth estimation networks involve dual [...] Read more.
Stereo-based depth estimation is becoming more and more important in many applications like self-driving vehicles, earth observation, cartography, robotics and so on. Modern approaches to depth estimation employ artificial intelligence techniques, particularly convolutional neural networks (CNNs). However, stereo-based depth estimation networks involve dual processing paths for left and right input images, which merge at intermediate layers, posing challenges for efficient deployment on modern hardware accelerators. Specifically, modern depth-first and layer-fused execution strategies, which are commonly used to reduce I/O communication and on-chip memory demands, are not readily compatible with such non-linear network structures. To address this limitation, we propose a flexible multi-core hardware architecture tailored for stereo-based depth estimation CNNs. The architecture supports layer-fused execution while efficiently managing dual-path computation and its fusion, enabling improved resource utilization. Experimental results demonstrate a latency reduction of up to 24% compared to state-of-the-art depth-first implementations that do not incorporate stereo-specific optimizations. Full article
(This article belongs to the Special Issue Multimedia Signal Processing and Computer Vision)
►▼ Show Figures

Figure 1

39 pages, 8097 KB  
Review
A Comprehensive Study on GaN Power Devices: Reliability, Performance, and Application Perspectives
by Susmita Mistri, Catherine Langpoklakpam, Surya Elangovan and Hao-Chung Kuo
Electronics 2025, 14(22), 4430; https://doi.org/10.3390/electronics14224430 - 13 Nov 2025
Cited by 13 | Viewed by 13332
Abstract
This review examines recent advances in Gallium Nitride (GaN) power semiconductor devices and their growing impact on the development of high-efficiency power conversion systems. It explores innovations in device design, packaging methods, and gate-driving strategies that have improved both performance and reliability. Key [...] Read more.
This review examines recent advances in Gallium Nitride (GaN) power semiconductor devices and their growing impact on the development of high-efficiency power conversion systems. It explores innovations in device design, packaging methods, and gate-driving strategies that have improved both performance and reliability. Key metrics such as switching speed, conduction losses, thermal management, and device robustness are analyzed, supported by reliability assessment techniques including Double-Pulse Testing (DPT). The discussion extends to current market dynamics and strategic industry initiatives that have catalyzed widespread GaN adoption. These combined insights highlight GaN’s role as a transformative material offering compact, efficient, and durable power solutions while identifying challenges that remain for broader implementation across diverse industries. Full article
(This article belongs to the Special Issue Advances in Semiconductor GaN and Applications)
►▼ Show Figures

Figure 1

19 pages, 2716 KB  
Article
Analysis of a Hybrid Intrabody Communications Scheme for Wireless Cortical Implants
by Assefa K. Teshome and Daniel T. H. Lai
Electronics 2025, 14(22), 4410; https://doi.org/10.3390/electronics14224410 - 12 Nov 2025
Viewed by 867
Abstract
Implantable technologies targeting the cerebral cortex and deeper brain structures are increasingly utilised in human–machine interfacing, advanced neuroprosthetics, and clinical interventions for neurological conditions. These systems require highly efficient and low-power methods for exchanging information between the implant and external electronics. Traditional approaches [...] Read more.
Implantable technologies targeting the cerebral cortex and deeper brain structures are increasingly utilised in human–machine interfacing, advanced neuroprosthetics, and clinical interventions for neurological conditions. These systems require highly efficient and low-power methods for exchanging information between the implant and external electronics. Traditional approaches often rely on inductively coupled data transfer (ic-DT), where the same coils used for wireless power are modulated for communication. Other designs use high-frequency antenna-based radio systems, typically operating in the 401–406 MHz MedRadio band or the 2.4 GHz ISM band. A promising alternative is intrabody communication (IBC), which leverages the bioelectrical characteristics of body tissue to enable signal propagation. This work presents a theoretical investigation into two schemes—inductive coupling and galvanically coupled IBC (gc-IBC)—as applied to cortical data links, considering frequencies from 1 to 10 MHz and implant depths of up to 7 cm. We propose a hybrid solution where gc-IBC supports data transmission and inductive coupling facilitates wireless power delivery. Our findings indicate that gc-IBC can accommodate wider bandwidths than ic-DT and offers significantly reduced path loss, approximately 20 dB lower than those of conventional RF-based antenna systems. Full article
(This article belongs to the Special Issue Applications of Sensor Networks and Wireless Communications)
►▼ Show Figures

Figure 1

10 pages, 2053 KB  
Article
A Terahertz Dual-Band Transmitter in 40 nm CMOS for a Wideband Sparse Synthetic Bandwidth Radar
by Aguan Hong, Lina Su, Yanjun Wang and Xiang Yi
Electronics 2025, 14(22), 4392; https://doi.org/10.3390/electronics14224392 - 11 Nov 2025
Viewed by 868
Abstract
This paper presents a terahertz (THz) dual-band transmitter for a wideband sparse synthetic bandwidth radar. The transmitter employs an innovative single-path-reuse dual-band architecture. This architecture utilizes a proposed quad-transformer-coupled voltage-controlled oscillator (VCO) as an on-chip local oscillator source. It also incorporates an innovative [...] Read more.
This paper presents a terahertz (THz) dual-band transmitter for a wideband sparse synthetic bandwidth radar. The transmitter employs an innovative single-path-reuse dual-band architecture. This architecture utilizes a proposed quad-transformer-coupled voltage-controlled oscillator (VCO) as an on-chip local oscillator source. It also incorporates an innovative dual-harmonic generator and a dual-band antenna, which work together within the single signal path to generate both the fundamental frequency and its second harmonic, thereby creating the dual bands required for a sparse synthetic bandwidth radar. Fabricated in a TSMC 40 nm CMOS technology, measurement results show that the transmitter achieves a peak equivalent isotropically radiated power (EIRP) of −7.95 dBm in the low-frequency band (121.34∼126.85 GHz) and −7.86 dBm in the high-frequency band (242.68∼253.7 GHz), validating the proposed architecture’s capability to generate dual-band signals simultaneously. The entire chip occupies a compact area of only 0.54 × 0.62 mm2 and consumes 136 mW of DC power. Full article
(This article belongs to the Special Issue Advancements in High-Performance RF Transceiver Technology: Innovations and Applications)
►▼ Show Figures

Figure 1

34 pages, 11286 KB  
Article
Degradation of Multi-Task Prompting Across Six NLP Tasks and LLM Families
by Federico Di Maio and Manuel Gozzi
Electronics 2025, 14(21), 4349; https://doi.org/10.3390/electronics14214349 - 6 Nov 2025
Cited by 2 | Viewed by 3261
Abstract
This study investigates how increasing prompt complexity affects the performance of Large Language Models (LLMs) across multiple Natural Language Processing (NLP) tasks. We introduce an incremental evaluation framework where six tasks—JSON formatting, English-Italian translation, sentiment analysis, emotion classification, topic extraction, and named entity [...] Read more.
This study investigates how increasing prompt complexity affects the performance of Large Language Models (LLMs) across multiple Natural Language Processing (NLP) tasks. We introduce an incremental evaluation framework where six tasks—JSON formatting, English-Italian translation, sentiment analysis, emotion classification, topic extraction, and named entity recognition—are progressively combined within a single prompt. Six representative open-source LLMs from different families (Llama 3.1 8B, Gemma 3 4B, Mistral 7B, Qwen3 4B, Granite 3.1 3B, and DeepSeek R1 7B) were systematically evaluated using local inference environments to ensure reproducibility. Results show that performance degradation is highly architecture-dependent: while Qwen3 4B maintained stable performance across all tasks, Gemma 3 4B and Granite 3.1 3B exhibited severe collapses in fine-grained semantic tasks. Interestingly, some models (e.g., Llama 3.1 8B and DeepSeek R1 7B) demonstrated positive transfer effects, improving in certain tasks under multitask conditions. Statistical analyses confirmed significant differences across models for structured and semantic tasks, highlighting the absence of a universal degradation rule. These findings suggest that multitask prompting resilience is shaped more by architectural design than by model size alone, and they motivate adaptive, model-specific strategies for prompt composition in complex NLP applications. Full article
(This article belongs to the Special Issue Artificial Intelligence-Driven Emerging Applications)
►▼ Show Figures

Figure 1

33 pages, 22059 KB  
Review
Resistive Sensing in Soft Robotic Grippers: A Comprehensive Review of Strain, Tactile, and Ionic Sensors
by Donya Mostaghniyazdi and Shahab Edin Nodehi
Electronics 2025, 14(21), 4290; https://doi.org/10.3390/electronics14214290 - 31 Oct 2025
Cited by 7 | Viewed by 5530
Abstract
Soft robotic grippers have emerged as crucial tools for safe and adaptive manipulation of delicate and different objects, enabled by their compliant structures. These grippers need embedded sensing that offers proprioceptive and exteroceptive feedback in order to function consistently. Resistive sensing is unique [...] Read more.
Soft robotic grippers have emerged as crucial tools for safe and adaptive manipulation of delicate and different objects, enabled by their compliant structures. These grippers need embedded sensing that offers proprioceptive and exteroceptive feedback in order to function consistently. Resistive sensing is unique among transduction processes since it is easy to use, scalable, and compatible with deformable materials. The three main classes of resistive sensors used in soft robotic grippers are systematically examined in this review: ionic sensors, which are emerging multimodal devices that can capture both mechanical and environmental cues; tactile sensors, which detect contact, pressure distribution, and slip; and strain sensors, which monitor deformation and actuation states. Their methods of operation, material systems, fabrication techniques, performance metrics, and integration plans are all compared in the survey. The results show that sensitivity, linearity, durability, and scalability are all trade-offs across sensor categories, with ionic sensors showing promise as a new development for multipurpose soft grippers. There is also a discussion of difficulties, including hysteresis, long-term stability, and signal processing complexity. In order to move resistive sensing from lab prototypes to reliable, practical applications in domains like healthcare, food handling, and human–robot collaboration, the review concludes that developments in hybrid material systems, additive manufacturing, and AI-enhanced signal interpretation will be crucial. Full article
(This article belongs to the Special Issue Advances in Environmental Sensing and Control for Mobile Robotics: Methods, Technologies, and Applications)
►▼ Show Figures

Figure 1

18 pages, 1539 KB  
Article
A Model of Output Power Control Method for Fault Ride-Through in a Single-Phase NPC Inverter-Based Power Conditioning System with IPOS DAB Converter and Battery
by Reo Emoto, Hiroaki Yamada and Tomokazu Mishima
Electronics 2025, 14(21), 4291; https://doi.org/10.3390/electronics14214291 - 31 Oct 2025
Viewed by 855
Abstract
Grid-tied inverters must satisfy fault ride-through (FRT) requirements to ensure grid stability during voltage disturbances. However, most existing FRT-related studies have focused on reactive current injection or voltage support functions, with few addressing how the active power reference should be dynamically controlled during [...] Read more.
Grid-tied inverters must satisfy fault ride-through (FRT) requirements to ensure grid stability during voltage disturbances. However, most existing FRT-related studies have focused on reactive current injection or voltage support functions, with few addressing how the active power reference should be dynamically controlled during voltage dips. In addition, few systems enable bidirectional power transfer or provide comprehensive verification under deep voltage dips. To address this issue, this paper proposes an output power control method for FRT in a single-phase neutral-point-clamped (NPC) inverter-based PCS consisting of an input-parallel output-series (IPOS) dual-active-bridge (DAB) converter and a battery. The proposed PCS dynamically reduces the output power reference according to the retained voltage while maintaining the inverter current within the rated limit, thereby ensuring stable operation. Computer simulations were conducted using Altair PSIM to verify the effectiveness of the proposed method. The results confirmed that the PCS satisfied the FRT requirements for all post-fault voltage levels. The injected current returned to its pre-fault value within 20 ms and 90 ms for 20% and 0% voltage dips, respectively, complying with the required recovery times. The proposed control method enhances grid resilience and maintains power quality in single-phase low-voltage distribution systems. Full article
(This article belongs to the Special Issue DC–DC Power Converter Technologies for Energy Storage Integration)
►▼ Show Figures

Graphical abstract

38 pages, 9535 KB  
Article
Novel Design and Experimental Validation of a Technique for Suppressing Distortion Originating from Various Sources in Multiantenna Full-Duplex Systems
by Keng-Hwa Liu, Juinn-Horng Deng and Min-Siou Yang
Electronics 2025, 14(21), 4300; https://doi.org/10.3390/electronics14214300 - 31 Oct 2025
Viewed by 809
Abstract
Complex distortion cancellation methods are often used at the radio frequency (RF) front end of multiantenna full-duplex transceivers to mitigate signal distortion; however, these methods have high computational complexity and limited practicality. To address these problems, the present study explored the complexities associated [...] Read more.
Complex distortion cancellation methods are often used at the radio frequency (RF) front end of multiantenna full-duplex transceivers to mitigate signal distortion; however, these methods have high computational complexity and limited practicality. To address these problems, the present study explored the complexities associated with such transceivers to develop a practical multistep approach for suppressing distortions arising from in-phase and quadrature (I/Q) imbalance, nonlinear power amplifier (PA) responses, and multipath self-interference caused by simultaneous transmissions on the same frequency. In this approach, the I/Q imbalance is estimated and then compensated for, following which nonlinear PA distortion is estimated and pre-compensated for. Subsequently, an auxiliary RF transmitter is combined with linearly regenerating self-interference signals to achieve full-duplex self-interference cancellation. The proposed method was implemented on a software-defined radio platform, with the distortion factor calibration specifically optimized for multiantenna full-duplex transceivers. The experimental results indicate that the image signal caused by I/Q imbalance can be suppressed by up to 60 dB through iterative computation. By combining IQI and DPD preprocessing, the nonlinear distortion spectrum can be reduced by 25 dB. Furthermore, integrating IQI, DPD, and self-interference preprocessing achieves up to 180 dB suppression of self-interference signals. Experimental results also demonstrate that the proposed method achieves approximately 20 dB suppression of self-interference. Thus, the method has high potential for enhancing the performance of multiantenna RF full-duplex systems. Full article
(This article belongs to the Section Microwave and Wireless Communications)
►▼ Show Figures

Figure 1

20 pages, 4224 KB  
Article
Reconfigurable Intelligence Surface Assisted Multiuser Downlink Communication with User Scheduling
by Zhengjun Dai and Xianyi Rui
Electronics 2025, 14(21), 4253; https://doi.org/10.3390/electronics14214253 - 30 Oct 2025
Viewed by 870
Abstract
The integration of Reconfigurable Intelligent Surfaces (RISs) into wireless networks is a promising paradigm for enhancing spectral efficiency and coverage in beyond-5G systems. However, in multiuser downlink scenarios, the joint optimization of discrete RIS phase shifts and user scheduling presents a high-dimensional combinatorial [...] Read more.
The integration of Reconfigurable Intelligent Surfaces (RISs) into wireless networks is a promising paradigm for enhancing spectral efficiency and coverage in beyond-5G systems. However, in multiuser downlink scenarios, the joint optimization of discrete RIS phase shifts and user scheduling presents a high-dimensional combinatorial challenge due to their tight coupling, which is often intractable with conventional methods. Furthermore, conventional RISs are limited by their unidirectional signal reflection, creating coverage blind spots. To address these issues, this paper first investigates a multi-user scheduling system assisted by a conventional RIS. We employed a vector projection relaxation method to transform the complex joint optimization problem, and then used an algorithm based on particle swarm optimization to jointly optimize the discrete phase shift and user scheduling. Simulation results demonstrate that this proposed algorithm significantly improves the system’s achievable data rate compared to existing benchmarks. Subsequently, to overcome the fundamental coverage limitation of conventional RISs, we extend our framework to two advanced systems: double-RIS and Simultaneously Transmitting and Reflecting RIS (STAR-RIS). For the STAR-RIS system, leveraging its energy-splitting protocol, we develop a novel joint optimization algorithm for phase shifts, amplitudes, and user scheduling based on an alternating optimization framework. This constitutes another significant contribution, as it effectively manages the added complexity of simultaneous transmission and reflection control. Simulations confirm that the STAR-RIS-assisted system, optimized by our algorithm, not only eliminates coverage blind spots but also surpasses the performance of traditional RIS, offering new perspectives for optimizing next-generation wireless communication networks. Full article
(This article belongs to the Section Microwave and Wireless Communications)
►▼ Show Figures

Figure 1

12 pages, 633 KB  
Article
Optimized FreeMark Post-Training White-Box Watermarking of Tiny Neural Networks
by Riccardo Adorante, Tullio Facchinetti and Danilo Pietro Pau
Electronics 2025, 14(21), 4237; https://doi.org/10.3390/electronics14214237 - 29 Oct 2025
Cited by 1 | Viewed by 774
Abstract
Neural networks are powerful, high-accuracy systems whose trained parameters represent a valuable intellectual property. Building models that reach top level performance is a complex task and requires substantial investments of time and money so protecting these assets is an increasingly important task. Extensive [...] Read more.
Neural networks are powerful, high-accuracy systems whose trained parameters represent a valuable intellectual property. Building models that reach top level performance is a complex task and requires substantial investments of time and money so protecting these assets is an increasingly important task. Extensive research has been carried out on Neural Network Watermarking, exploring the possibility of inserting a recognizable marker in a host model either in the form of a concealed bit-string or as a characteristic output, making it possible to confirm network ownership even in the presence of malicious attempts at erasing the embedded marker from the model. The study examines the applicability of Opt-FreeMark, a non-invasive post-training white-box watermarking technique, obtained by modifying and optimizing an already existing state-of-the-art technique for tiny neural networks. Here, “Tiny” refers to models intended for ultra-low-power deployments, such as those running on edge devices like sensors and micro-controllers. Watermark robustness is also demonstrated by simulating common model-modification attacks that try to eliminate it from the model while preserving performance; the results presented in the paper indicate that the watermarking scheme effectively protects the networks against these manipulations. Full article
(This article belongs to the Special Issue Machine Learning in Electronic and Biomedical Engineering, 4th Edition)
►▼ Show Figures

Figure 1

26 pages, 3558 KB  
Article
Avocado: An Interpretable Fine-Grained Intrusion Detection Model for Advanced Industrial Control Network Attacks
by Xin Liu, Tao Liu and Ning Hu
Electronics 2025, 14(21), 4233; https://doi.org/10.3390/electronics14214233 - 29 Oct 2025
Viewed by 1073
Abstract
Industrial control systems (ICS), as critical infrastructure supporting national operations, are increasingly threatened by sophisticated stealthy network attacks. These attacks often break malicious behaviors into multiple highly camouflaged packets, which are embedded into large-scale background traffic with low frequency, making them semantically and [...] Read more.
Industrial control systems (ICS), as critical infrastructure supporting national operations, are increasingly threatened by sophisticated stealthy network attacks. These attacks often break malicious behaviors into multiple highly camouflaged packets, which are embedded into large-scale background traffic with low frequency, making them semantically and temporally indistinguishable from normal traffic and thus evading traditional detection. Existing methods largely rely on flow-level statistics or long-sequence modeling, resulting in coarse detection granularity, high latency, and poor byte-level interpretability, falling short of industrial demands for real-time and actionable detection. To address these challenges, we propose Avocado, a fine-grained, multi-level intrusion detection model. Avocado’s core innovation lies in contextual flow-feature fusion: it models each packet jointly with its surrounding packet sequence, enabling independent abnormality detection and precise localization. Moreover, a shared-query multi-head self-attention mechanism is designed to quantify byte-level importance within packets. Experimental results show that Avocado significantly outperforms state-of-the-art flow-level methods on NGAS and CLIA-M221 datasets, improving packet-level detection ACC by 1.55% on average, and reducing FPR and FNR to 3.2%, 3.6% (NGAS), and 3.7%, 4.3% (CLIA-M221), respectively, demonstrating its superior performance in both detection and interpretability. Full article
(This article belongs to the Special Issue Novel Approaches for Deep Learning in Cybersecurity)
►▼ Show Figures

Figure 1

19 pages, 4017 KB  
Article
LACX: Locality-Aware Shared Data Migration in NUMA + CXL Tiered Memory
by Hayong Jeong, Binwon Song, Minwoo Jo and Heeseung Jo
Electronics 2025, 14(21), 4235; https://doi.org/10.3390/electronics14214235 - 29 Oct 2025
Viewed by 1525
Abstract
In modern high-performance computing (HPC) and large-scale data processing environments, the efficient utilization and scalability of memory resources are critical determinants of overall system performance. Architectures such as non-uniform memory access (NUMA) and tiered memory systems frequently suffer performance degradation due to remote [...] Read more.
In modern high-performance computing (HPC) and large-scale data processing environments, the efficient utilization and scalability of memory resources are critical determinants of overall system performance. Architectures such as non-uniform memory access (NUMA) and tiered memory systems frequently suffer performance degradation due to remote accesses stemming from shared data among multiple tasks. This paper proposes LACX, a shared data migration technique leveraging Compute Express Link (CXL), to address these challenges. LACX preserves the migration cycle of automatic NUMA balancing (AutoNUMA) while identifying shared data characteristics and migrating such data to CXL memory instead of DRAM, thereby maximizing DRAM locality. The proposed method utilizes existing kernel structures and data to efficiently identify and manage shared data without incurring additional overhead, and it effectively avoids conflicts with AutoNUMA policies. Evaluation results demonstrate that, although remote accesses to shared data can degrade performance in low-tier memory scenarios, LACX significantly improves overall memory bandwidth utilization and system performance in high-tier memory and memory-intensive workload environments by distributing DRAM bandwidth. This work presents a practical, lightweight approach to shared data management in tiered memory environments and highlights new directions for next-generation memory management policies. Full article
(This article belongs to the Special Issue Future Technologies for Data Management, Processing and Application)
►▼ Show Figures

Figure 1

38 pages, 9358 KB  
Article
Generation of a Multi-Class IoT Malware Dataset for Cybersecurity
by Mazdak Maghanaki, Soraya Keramati, F. Frank Chen and Mohammad Shahin
Electronics 2025, 14(21), 4196; https://doi.org/10.3390/electronics14214196 - 27 Oct 2025
Cited by 10 | Viewed by 3097
Abstract
This study introduces a modular, behaviorally curated malware dataset suite consisting of eight independent sets, each specifically designed to represent a single malware class: Trojan, Mirai (botnet), ransomware, rootkit, worm, spyware, keylogger, and virus. In contrast to earlier approaches that aggregate all malware [...] Read more.
This study introduces a modular, behaviorally curated malware dataset suite consisting of eight independent sets, each specifically designed to represent a single malware class: Trojan, Mirai (botnet), ransomware, rootkit, worm, spyware, keylogger, and virus. In contrast to earlier approaches that aggregate all malware into large, monolithic collections, this work emphasizes the selection of features unique to each malware type. Feature selection was guided by established domain knowledge and detailed behavioral telemetry obtained through sandbox execution and a subsequent report analysis on the AnyRun platform. The datasets were compiled from two primary sources: (i) the AnyRun platform, which hosts more than two million samples and provides controlled, instrumented sandbox execution for malware, and (ii) publicly available GitHub repositories. To ensure data integrity and prevent cross-contamination of behavioral logs, each sample was executed in complete isolation, allowing for the precise capture of both static attributes and dynamic runtime behavior. Feature construction was informed by operational signatures characteristic of each malware category, ensuring that the datasets accurately represent the tactics, techniques, and procedures distinguishing one class from another. This targeted design enabled the identification of subtle but significant behavioral markers that are frequently overlooked in aggregated datasets. Each dataset was balanced to include benign, suspicious, and malicious samples, thereby supporting the training and evaluation of machine learning models while minimizing bias from disproportionate class representation. Across the full suite, 10,000 samples and 171 carefully curated features were included. This constitutes one of the first dataset collections intentionally developed to capture the behavioral diversity of multiple malware categories within the context of Internet of Things (IoT) security, representing a deliberate effort to bridge the gap between generalized malware corpora and class-specific behavioral modeling. Full article
(This article belongs to the Special Issue Machine Learning for Cyber Security and Privacy: Innovations, Challenges, and Future Directions)
►▼ Show Figures

Graphical abstract

25 pages, 22171 KB  
Article
Physics-Informed Co-Optimization of Fuel-CellFlying Vehicle Propulsion and Control Systems with Onboard Catalysis
by Yifei Bao, Chaoyi Chen, Hao Zhang and Nuo Lei
Electronics 2025, 14(21), 4150; https://doi.org/10.3390/electronics14214150 - 23 Oct 2025
Viewed by 993
Abstract
Fuel-cell flying vehicles suffer from limited endurance, while ammonia, decomposed onboard to supply hydrogen, offers a carbon-free, high-density solution to extend flight missions. However, the system’s performance is governed by a multi-scale coupling between propulsion and control systems. To this end, this paper [...] Read more.
Fuel-cell flying vehicles suffer from limited endurance, while ammonia, decomposed onboard to supply hydrogen, offers a carbon-free, high-density solution to extend flight missions. However, the system’s performance is governed by a multi-scale coupling between propulsion and control systems. To this end, this paper introduces a novel optimization paradigm, termed physics-informed gradient-enhanced multi-objective optimization (PI-GEMO), to simultaneously optimize the ammonia decomposition unit (ADU) catalyst composition, powertrain sizing, and flight control parameters. The PI-GEMO framework leverages a physics-informed neural network (PINN) as a differentiable surrogate model, which is trained not only on sparse simulation data but also on the governing differential equations of the system. This enables the use of analytical gradient information extracted from the trained PINN via automatic differentiation to intelligently guide the evolutionary search process. A comprehensive case study on a flying vehicle demonstrates that the PI-GEMO framework not only discovers a superior set of Pareto-optimal solutions compared to traditional methods but also critically ensures the physical plausibility of the results. Full article
(This article belongs to the Special Issue Eco-Safe Intelligent Mobility Development and Application)
►▼ Show Figures

Figure 1

Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying article 151-200 on page 4 of 21.
Back to TopTop