Electronics

21 pages, 10373 KiB

Open AccessArticle

A 75 kW Medium-Frequency Transformer Design Based in Inductive Power Transfer (IPT) for Medium-Voltage Solid-State Transformer Applications

by Juan Blanco-Ortiz, Eduardo García-Martínez, Ignacio González-Prieto and Mario J. Duran

Electronics 2025, 14(6), 1059; https://doi.org/10.3390/electronics14061059 (registering DOI) - 7 Mar 2025

Abstract

Solid-State Transformers (SSTs) enable significant improvements in size and functionality compared to conventional power transformers. However, one of the key challenges in Solid-State Transformer design is achieving reliable insulation between the high-voltage and low-voltage sections. This proposal presents the design and optimization of [...] Read more.

Solid-State Transformers (SSTs) enable significant improvements in size and functionality compared to conventional power transformers. However, one of the key challenges in Solid-State Transformer design is achieving reliable insulation between the high-voltage and low-voltage sections. This proposal presents the design and optimization of a high-insulation Medium-Frequency Transformer (MFT) for 66 kV grids operating at 50 kHz and delivering up to 75 kW for SST applications using Inductive Power Transfer (IPT) technology. A fixed 50 mm gap between the primary and secondary windings is filled with dielectric oil to enhance insulation. The proposed IPT system employs a double-D coil design developed through iterative 2D and 3D finite element method simulations to optimize the magnetic circuit, thereby significantly reducing stray flux and losses. Notably, the double-D configuration reduces enclosure losses from 269.6 W, observed in a rectangular coil design, to 4.38 W, resulting in an overall system loss reduction of 42.4% while maintaining the electrical parameters required for zero-voltage switching operation. These advancements address the critical limitations in conventional Medium-Frequency Transformers by providing enhanced insulation and improved thermal management. The proposed IPT-based design offers a low-loss solution with easy thermal management for solid-state transformer applications in high-voltage grids. Full article

(This article belongs to the Special Issue Advanced Control Techniques for Power Electronics: Addressing Challenges in Renewable Energy System Applications)

► Show Figures

Figure 1

15 pages, 3702 KiB

Open AccessArticle

Multiple Differential Convolution and Local-Variation Attention UNet: Nucleus Semantic Segmentation Based on Multiple Differential Convolution and Local-Variation Attention

by Xiaoming Sun, Shilin Li, Yongji Chen, Junxia Chen, Hao Geng, Kun Sun, Yuemin Zhu, Bochao Su and Hu Zhang

Electronics 2025, 14(6), 1058; https://doi.org/10.3390/electronics14061058 (registering DOI) - 7 Mar 2025

Abstract

Nucleus accurate segmentation is a crucial task in biomedical image analysis. While convolutional neural networks (CNNs) have achieved notable progress in this field, challenges remain due to the complexity and heterogeneity of cell images, especially in overlapping regions of nuclei. To address the [...] Read more.

Nucleus accurate segmentation is a crucial task in biomedical image analysis. While convolutional neural networks (CNNs) have achieved notable progress in this field, challenges remain due to the complexity and heterogeneity of cell images, especially in overlapping regions of nuclei. To address the limitations of current methods, we propose a mechanism of multiple differential convolution and local-variation attention in CNNs, leading to the so-called multiple differential convolution and local-variation attention U-Net (MDLA-UNet). The multiple differential convolution employs multiple differential operators to capture gradient and direction information, improving the network’s capability to detect edges. The local-variation attention utilizes Haar discrete wavelet transforms for level-1 decomposition to obtain approximate features, and then derives high-frequency features to enhance the global context and local detail variation of the feature maps. The results on the MoNuSeg, TNBC, and CryoNuSeg datasets demonstrated superior segmentation performance of the proposed method for cells having complex boundaries and details with respect to existing methods. The proposed MDLA-UNet presents the ability of capturing fine edges and details in feature maps and thus improves the segmentation of nuclei with blurred boundaries and overlapping regions. Full article

(This article belongs to the Special Issue Feature Papers in "Computer Science & Engineering", 2nd Edition)

► Show Figures

Graphical abstract

22 pages, 4433 KiB

Open AccessArticle

FCCA: Fast Center Consistency Attention for Facial Expression Recognition

by Rui Sun, Zhaoli Zhang and Hai Liu

Electronics 2025, 14(6), 1057; https://doi.org/10.3390/electronics14061057 (registering DOI) - 7 Mar 2025

Abstract

Given the critical requirements for both speed and accuracy in facial expression recognition, this paper presents a novel deep-learning architecture named Fast Central Consistency Attention (FCCA). With FasterNet-s as its backbone network, FCCA is designed to recognize facial expressions. Firstly, we leverage partial [...] Read more.

Given the critical requirements for both speed and accuracy in facial expression recognition, this paper presents a novel deep-learning architecture named Fast Central Consistency Attention (FCCA). With FasterNet-s as its backbone network, FCCA is designed to recognize facial expressions. Firstly, we leverage partial convolution to extract features from specific channels, thereby reducing frequent memory access and substantially boosting training speed. Secondly, we enhance recognition accuracy by introducing an additional pointwise convolution on the partial features, focusing on the central facial position using weighted mechanisms. Lastly, we integrate flip consistency loss to tackle uncertainty challenges inherent in facial expression recognition (FER) tasks, further improving the overall model performance. Our approach yielded superior results: we achieved recognition accuracies of 91.30% on RAF-DB and 65.51% on AffectNet datasets, along with 56.61% UAR and 69.66% WAR on the DFEW dataset. The FCCA method has demonstrated state-of-the-art performance across multiple datasets, underscoring its robustness and capability for generalization. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

19 pages, 3943 KiB

Open AccessArticle

Quad-Beam 4 × 2 Array Antenna for Millimeter-Wave 5G Applications

by Parveez Shariff Bhadravathi Ghouse, Tanweer Ali, Pallavi R. Mane, Sameena Pathan, Sudheesh Puthenveettil Gopi, Bal S. Virdee, Jaume Anguera and Prashant M. Prabhu

Electronics 2025, 14(5), 1056; https://doi.org/10.3390/electronics14051056 (registering DOI) - 6 Mar 2025

Abstract

This article presents the design of a novel, compact, 4 × 2 planar-array antenna that provides quad-beam radiation in the broadside direction, and it enhances coverage and serviceability for millimeter-wave applications. The antenna utilizes a corporate (parallel) feed network to deliver equal power [...] Read more.

This article presents the design of a novel, compact, 4 × 2 planar-array antenna that provides quad-beam radiation in the broadside direction, and it enhances coverage and serviceability for millimeter-wave applications. The antenna utilizes a corporate (parallel) feed network to deliver equal power and phase to all elements. Non-uniform element spacing in the two orthogonal planes, exceeding 0.5

λ_{1}

(

λ_{1}

being the wavelength at 30 GHz), results in a quad-beam radiation pattern. Two beams are formed in the xz-plane and two in the yz-plane, oriented at angles of

θ = \pm 54 °

. However, this spacing leads to null radiation at the center and splits the radiation energy, reducing the overall gain. The measured half-power beamwidth (HPBW) is 30° in the xz-plane and 35° in the yz-plane, with X-polarization levels of −20.5 dB and −26 dB, respectively. The antenna achieves a bandwidth of 28.5–31.1 GHz and a peak gain of 10.6 dBi. Furthermore, increasing the aperture size enhances the gain and narrows the beamwidth by replicating the structure and tuning the feed network. These features make the proposed antenna suitable for 5G wireless communication systems. Full article

► Show Figures

Figure 1

22 pages, 1744 KiB

Open AccessArticle

Hybrid Long-Range–5G Multi-Sensor Platform for Predictive Maintenance for Ventilation Systems

by Praveen Mohanram and Robert H. Schmitt

Electronics 2025, 14(5), 1055; https://doi.org/10.3390/electronics14051055 - 6 Mar 2025

Abstract

In this paper, we present a multi-sensor platform for predictive maintenance featuring hybrid long-range (LoRa) and 5G connectivity. This hybrid approach combines LoRa’s low-power transmission for energy efficiency with 5G’s real-time data capabilities. The hardware platform integrates multiple sensors to monitor machine health [...] Read more.

In this paper, we present a multi-sensor platform for predictive maintenance featuring hybrid long-range (LoRa) and 5G connectivity. This hybrid approach combines LoRa’s low-power transmission for energy efficiency with 5G’s real-time data capabilities. The hardware platform integrates multiple sensors to monitor machine health parameters, with data analyzed on the device using pre-trained AI models to assess the machine’s condition. Inferences are transmitted via LoRa to the operator for maintenance scheduling, while a cloud application tracks and stores sensor data. Periodic sensor data bursts are sent via 5G to update the AI model, which is then delivered back to the platform through over-the-air (OTA) updates. We provide a comprehensive overview of the hardware architecture, along with an in-depth analysis of the data generated by the sensors, and its processing methodology. However, the data analysis and the software for ventilation control and its predictive capabilities are not the focus of this paper and are not presented. Full article

(This article belongs to the Special Issue 5G Mobile Telecommunication Systems and Recent Advances)

► Show Figures

Figure 1

19 pages, 9660 KiB

Open AccessArticle

An Efficient Synthetic Aperture Radar Interference Suppression Method Based on Image Domain Regularization

by Xuyang Ge, Xingdong Liang, Hang Li, Zhiyu Jiang, Yuan Zhang and Xiangxi Bu

Electronics 2025, 14(5), 1054; https://doi.org/10.3390/electronics14051054 (registering DOI) - 6 Mar 2025

Abstract

Synthetic aperture radar (SAR) systems, as wideband radar systems, are inherently susceptible to interference signals within their operational frequency band, which significantly affects SAR signal processing and image interpretation. Recent studies have demonstrated that semiparametric methods (e.g., the RPCA method) exhibit excellent performance [...] Read more.

Synthetic aperture radar (SAR) systems, as wideband radar systems, are inherently susceptible to interference signals within their operational frequency band, which significantly affects SAR signal processing and image interpretation. Recent studies have demonstrated that semiparametric methods (e.g., the RPCA method) exhibit excellent performance in suppressing these interference signals. However, these methods predominantly focus on processing SAR’s raw echo data, which does not satisfy the sparsity requirements and entails extremely high computational complexity, complicating integration with imaging algorithms. This paper introduces an effective method for suppressing interference signals by leveraging the sparsity of the SAR image domain. It utilizes the sparsity of the interference signal in the two-dimensional frequency domain, following focusing processing, rather than relying on low-rank properties. This approach significantly reduces the computational complexity. Ultimately, the effectiveness and efficiency of the proposed algorithm are validated through experiments conducted with simulated and real SAR data. Full article

(This article belongs to the Special Issue New Challenges in Remote Sensing Image Processing)

► Show Figures

Figure 1

26 pages, 6237 KiB

Open AccessArticle

Generative AI in Education: Perspectives Through an Academic Lens

by Iulian Întorsureanu, Simona-Vasilica Oprea, Adela Bâra and Dragoș Vespan

Electronics 2025, 14(5), 1053; https://doi.org/10.3390/electronics14051053 - 6 Mar 2025

Abstract

In this paper, we investigated the role of generative AI in education in academic publications extracted from Web of Science (3506 records; 2019–2024). The proposed methodology included three main streams: (1) Monthly analysis trends; top-ranking research areas, keywords and universities; frequency of keywords [...] Read more.

In this paper, we investigated the role of generative AI in education in academic publications extracted from Web of Science (3506 records; 2019–2024). The proposed methodology included three main streams: (1) Monthly analysis trends; top-ranking research areas, keywords and universities; frequency of keywords over time; a keyword co-occurrence map; collaboration networks; and a Sankey diagram illustrating the relationship between AI-related terms, publication years and research areas; (2) Sentiment analysis using a custom list of words, VADER and TextBlob; (3) Topic modeling using Latent Dirichlet Allocation (LDA). Terms such as “artificial intelligence” and “generative artificial intelligence” were predominant, but they diverged and evolved over time. By 2024, AI applications had branched into specialized fields, including education and educational research, computer science, engineering, psychology, medical informatics, healthcare sciences, general medicine and surgery. The sentiment analysis reveals a growing optimism in academic publications regarding generative AI in education, with a steady increase in positive sentiment from 2023 to 2024, while maintaining a predominantly neutral tone. Five main topics were derived from AI applications in education, based on an analysis of the most relevant terms extracted by LDA: (1) Gen-AI’s impact in education and research; (2) ChatGPT as a tool for university students and teachers; (3) Large language models (LLMs) and prompting in computing education; (4) Applications of ChatGPT in patient education; (5) ChatGPT’s performance in medical examinations. The research identified several emerging topics: discipline-specific application of LLMs, multimodal gen-AI, personalized learning, AI as a peer or tutor and cross-cultural and multilingual tools aimed at developing culturally relevant educational content and supporting the teaching of lesser-known languages. Further, gamification with generative AI involves designing interactive storytelling and adaptive educational games to enhance engagement and hybrid human–AI classrooms explore co-teaching dynamics, teacher–student relationships and the impact on classroom authority. Full article

(This article belongs to the Special Issue Techniques and Applications in Prompt Engineering and Generative AI)

► Show Figures

Figure 1

22 pages, 3393 KiB

Open AccessArticle

A Dynamic Spatio-Temporal Traffic Prediction Model Applicable to Low Earth Orbit Satellite Constellations

by Kexuan Liu, Yasheng Zhang and Shan Lu

Electronics 2025, 14(5), 1052; https://doi.org/10.3390/electronics14051052 - 6 Mar 2025

Abstract

Low Earth Orbit (LEO) constellations support the transmission of various communication services and have been widely applied in fields such as global Internet access, the Internet of Things, remote sensing monitoring, and emergency communication. With the surge in traffic volume, the quality of [...] Read more.

Low Earth Orbit (LEO) constellations support the transmission of various communication services and have been widely applied in fields such as global Internet access, the Internet of Things, remote sensing monitoring, and emergency communication. With the surge in traffic volume, the quality of user services has faced unprecedented challenges. Achieving accurate low Earth orbit constellation network traffic prediction can optimize resource allocation, enhance the performance of LEO constellation networks, reduce unnecessary costs in operation management, and enable the system to adapt to the development of future services. Ground networks often adopt methods such as machine learning (support vector machine, SVM) or deep learning (convolutional neural network, CNN; generative adversarial network, GAN) to predict future short- and long-term traffic information, aiming to optimize network performance and ensure service quality. However, these methods lack an understanding of the high-dynamics of LEO satellites and are not applicable to LEO constellations. Therefore, designing an intelligent traffic prediction model that can accurately predict multi-service scenarios in LEO constellations remains an unsolved challenge. In this paper, in light of the characteristics of high-dynamics and the high-frequency data streams of LEO constellation traffic, the authors propose a DST-LEO satellite-traffic prediction model (a dynamic spatio-temporal low Earth orbit satellite traffic prediction model). This model captures the implicit features among satellite nodes through multiple attention mechanism modules and processes the traffic volume and traffic connection/disconnection data of inter-satellite links via a multi-source data separation and fusion strategy, respectively. After splicing and fusing at a specific scale, the model performs prediction through the attention mechanism. The model proposed by the authors achieved a short-term prediction RMSE of 0.0028 and an MAE of 0.0018 on the Abilene dataset. For long-term prediction on the Abilene dataset, the RMSE was 0.0054 and the MAE was 0.0039. The RMSE of the short-term prediction on the dataset simulated by the internal low Earth orbit constellation business simulation system was 0.0034, and the MAE was 0.0026. For the long-term prediction, the RMSE reached 0.0029 and the MAE reached 0.0022. Compared with other time series prediction models, it decreased by 22.3% in terms of the mean squared error and 18.0% in terms of the mean absolute error. The authors validated the functions of each module within the model through ablation experiments and further analyzed the effectiveness of this model in the task of LEO constellation network traffic prediction. Full article

(This article belongs to the Special Issue Future Generation Non-Terrestrial Networks)

► Show Figures

Figure 1

12 pages, 4424 KiB

Open AccessArticle

Self-Calibration Method for Accurate Direct-Current Ratio Calibration

by Changwei Zhai, Jianting Zhao, Yunfeng Lu, Pengcheng Hu and Kunli Zhou

Electronics 2025, 14(5), 1051; https://doi.org/10.3390/electronics14051051 - 6 Mar 2025

Abstract

A self-calibration method is proposed for determining the current ratio for DC current ratio devices. To demonstrate the feasibility of the proposal, an experimental test with nominal current ratios of 10:1 100:1 and 1000:1 is performed, and a three-step calibration procedure is used: [...] Read more.

A self-calibration method is proposed for determining the current ratio for DC current ratio devices. To demonstrate the feasibility of the proposal, an experimental test with nominal current ratios of 10:1 100:1 and 1000:1 is performed, and a three-step calibration procedure is used: Firstly, eleven magnetic modulator-based DC current transducers (DCCTs) with a nominal current ratio of 10:1 are used to realize the self-calibration of a 10:1 current ratio through the reference current method; secondly, the eleven DCCTs are series-parallel to achieve the current ratio build-up from 10:1 to 100:1; finally, another DCCT with a 100:1 auxiliary current ratio is used to realize the 1000:1 current ratio calibration, combining the two steps before. The merit of the proposed self-calibration method is that the measurement does not need a high-precision current source or ammeter. Current ratio calibration based on this technology offers a lower-cost, more robust alternative to those using other devices. The experimental results show that the uncertainty of the calibration, mainly dominated by the noise and the stability of the magnetic modulator used, is within a few parts in 10⁻⁷ or better in the 1000 A range. Full article

► Show Figures

Figure 1

13 pages, 3791 KiB

Open AccessArticle

γ-Fe₂O₃-Based MEMS Gas Sensor for Propane Detection

by Xiang Gao, Ying Chen, Pengcheng Xu, Dan Zheng and Xinxin Li

Electronics 2025, 14(5), 1050; https://doi.org/10.3390/electronics14051050 - 6 Mar 2025

Abstract

The selective detection of propane gas molecules using semiconductor gas sensors has always been a challenge within research. In this study, we successfully synthesized a γ-Fe₂O₃ nanomaterial with a selective catalytic effect on propane and loaded it onto a ZnO [...] Read more.

The selective detection of propane gas molecules using semiconductor gas sensors has always been a challenge within research. In this study, we successfully synthesized a γ-Fe₂O₃ nanomaterial with a selective catalytic effect on propane and loaded it onto a ZnO sensing material to construct a double-layer microsensor, which showed good sensing response characteristics in the detection of the refrigerant R290 (which is mainly propane). In addition, we also prepared a series of iron oxides, including nanomaterials such as α-Fe₂O₃, Fe₃O₄, and FeO, as well as γ-Fe₂O₃ materials with different specific surface areas obtained at various processing temperatures, and we carried out gas sensing research on R290. The results show that the γ-Fe₂O₃ material has a better sensitivity to R290, and the γ-Fe₂O₃ material calcined at 200 °C shows the best performance. Our results can provide a theoretical basis for the design and optimization of semiconductor gas sensors for alkane detection. Full article

(This article belongs to the Special Issue High-Performance Gas Sensors: Materials, Technologies and Applications)

► Show Figures

Figure 1

11 pages, 4419 KiB

Open AccessArticle

Investigation of Torque Ripple in Servo Motors with Different Magnet Geometries

by Hacı Dedecan, Engin Ayçiçek and Mustafa Gürkan Aydeniz

Electronics 2025, 14(5), 1049; https://doi.org/10.3390/electronics14051049 - 6 Mar 2025

Abstract

Servo motors are among the most efficient and precise performers within the category of permanent magnet synchronous motors. These motors stand out for their high power density, quiet operation, low maintenance, and wide operating speed range advantages. One of the disadvantages of these [...] Read more.

Servo motors are among the most efficient and precise performers within the category of permanent magnet synchronous motors. These motors stand out for their high power density, quiet operation, low maintenance, and wide operating speed range advantages. One of the disadvantages of these motors, which is also the subject of this study, is their high torque ripple. Torque ripple is critical in applications requiring precision, as it can affect operational performance and contribute to vibration and noise issues. Torque ripple can be reduced through design methods such as different winding layouts, slot openings, stator/rotor skewing, or pole offset. In this study, torque ripple of servo motors was investigated through various magnet geometry designs and analyses using the finite element method. Design and analysis studies were conducted for a reference servo motor, and alternative designs were obtained by modifying the rotor structure of the reference motor. In the studies conducted, it has been observed that the torque ripple, initially at 2.17 Nm, can be improved to as low as 1.23 Nm. This indicates that the torque ripple, which was initially at 3.75%, can be reduced to around 2.08%. However, performance losses may occur depending on the extent of improvement. Full article

(This article belongs to the Special Issue Advanced Design in Electrical Machines)

► Show Figures

Figure 1

30 pages, 3046 KiB

Open AccessReview

A Survey of Advancements in Scheduling Techniques for Efficient Deep Learning Computations on GPUs

by Rupinder Kaur, Arghavan Asad, Seham Al Abdul Wahid and Farah Mohammadi

Electronics 2025, 14(5), 1048; https://doi.org/10.3390/electronics14051048 - 6 Mar 2025

Abstract

This comprehensive survey explores recent advancements in scheduling techniques for efficient deep learning computations on GPUs. The article highlights challenges related to parallel thread execution, resource utilization, and memory latency in GPUs, which can lead to suboptimal performance. The surveyed research focuses on [...] Read more.

This comprehensive survey explores recent advancements in scheduling techniques for efficient deep learning computations on GPUs. The article highlights challenges related to parallel thread execution, resource utilization, and memory latency in GPUs, which can lead to suboptimal performance. The surveyed research focuses on novel scheduling policies to improve memory latency tolerance, exploit parallelism, and enhance GPU resource utilization. Additionally, it explores the integration of prefetching mechanisms, fine-grained warp scheduling, and warp switching strategies to optimize deep learning computations. These techniques demonstrate significant improvements in throughput, memory bank parallelism, and latency reduction. The insights gained from this survey can guide researchers, system designers, and practitioners in developing more efficient and powerful deep learning systems on GPUs. Furthermore, potential future research directions include advanced scheduling techniques, energy efficiency considerations, and the integration of emerging computing technologies. Through continuous advancement in scheduling techniques, the full potential of GPUs can be unlocked for a wide range of applications and domains, including GPU-accelerated deep learning, task scheduling, resource management, memory optimization, and more. Full article

(This article belongs to the Special Issue Emerging Applications of FPGAs and Reconfigurable Computing System)

► Show Figures

Figure 1

16 pages, 716 KiB

Open AccessArticle

Efficient Graph Representation Learning by Non-Local Information Exchange

by Ziquan Wei, Tingting Dan, Jiaqi Ding and Guorong Wu

Electronics 2025, 14(5), 1047; https://doi.org/10.3390/electronics14051047 - 6 Mar 2025

Abstract

Graphs are an effective data structure for characterizing ubiquitous connections as well as evolving behaviors that emerge in inter-wined systems. Limited by the stereotype of node-to-node connections, learning node representations is often confined in a graph diffusion process where local information has been [...] Read more.

Graphs are an effective data structure for characterizing ubiquitous connections as well as evolving behaviors that emerge in inter-wined systems. Limited by the stereotype of node-to-node connections, learning node representations is often confined in a graph diffusion process where local information has been excessively aggregated, as the random walk of graph neural networks (GNN) explores far-reaching neighborhoods layer-by-layer. In this regard, tremendous efforts have been made to alleviate feature over-smoothing issues such that current backbones can lend themselves to be used in a deep network architecture. However, compared to designing a new GNN, less attention has been paid to underlying topology by graph re-wiring, which mitigates not only flaws of the random walk but also the over-smoothing risk incurred by reducing unnecessary diffusion in deep layers. Inspired by the notion of non-local mean techniques in the area of image processing, we propose a non-local information exchange mechanism by establishing an express connection to the distant node, instead of propagating information along the (possibly very long) original pathway node-after-node. Since the process of seeking express connections throughout a graph can be computationally expensive in real-world applications, we propose a re-wiring framework (coined the express messenger wrapper) to progressively incorporate express links in a non-local manner, which allows us to capture multi-scale features without using a very deep model; our approach is thus free of the over-smoothing challenge. We integrate our express messenger wrapper with existing GNN backbones (either using graph convolution or tokenized transformer) and achieve a new record on the Roman-empire dataset as well as in terms of SOTA performance on both homophilous and heterophilous datasets. Full article

(This article belongs to the Special Issue Artificial Intelligence in Graphics and Images)

► Show Figures

Figure 1

23 pages, 7710 KiB

Open AccessArticle

Immersive Interaction for Inclusive Virtual Reality Navigation: Enhancing Accessibility for Socially Underprivileged Users

by Jeonghyeon Kim, Jung-Hoon Ahn and Youngwon Kim

Electronics 2025, 14(5), 1046; https://doi.org/10.3390/electronics14051046 - 6 Mar 2025

Abstract

Existing virtual reality (VR) street view and 360-degree road view applications often rely on complex controllers or touch interfaces, which can hinder user immersion and accessibility. These challenges are particularly pronounced for under-represented populations, such as older adults and individuals with limited familiarity [...] Read more.

Existing virtual reality (VR) street view and 360-degree road view applications often rely on complex controllers or touch interfaces, which can hinder user immersion and accessibility. These challenges are particularly pronounced for under-represented populations, such as older adults and individuals with limited familiarity with digital devices. Such groups frequently face physical or environmental constraints that restrict their ability to engage in outdoor activities, highlighting the need for alternative methods of experiencing the world through virtual environments. To address this issue, we propose a VR street view application featuring an intuitive, gesture-based interface designed to simplify user interaction and enhance accessibility for socially disadvantaged individuals. Our approach seeks to optimize digital accessibility by reducing barriers to entry, increasing user immersion, and facilitating a more inclusive virtual exploration experience. Through usability testing and iterative design, this study evaluates the effectiveness of gesture-based interactions in improving accessibility and engagement. The findings emphasize the importance of user-centered design in fostering an inclusive VR environment that accommodates diverse needs and abilities. Full article

► Show Figures

Figure 1

31 pages, 875 KiB

Open AccessArticle

Hierarchical Traffic Engineering in 3D Networks Using QoS-Aware Graph-Based Deep Reinforcement Learning

by Robert Kołakowski, Lechosław Tomaszewski, Rafał Tępiński and Sławomir Kukliński

Electronics 2025, 14(5), 1045; https://doi.org/10.3390/electronics14051045 - 6 Mar 2025

Abstract

Ubiquitous connectivity is envisioned through the integration of terrestrial (TNs) and non-terrestrial networks (NTNs). However, NTNs face multiple routing and Quality of Service (QoS) provisioning challenges due to the mobility of network nodes. Distributed Software-Defined Networking (SDN) combined with Multi-Agent Deep Reinforcement Learning [...] Read more.

Ubiquitous connectivity is envisioned through the integration of terrestrial (TNs) and non-terrestrial networks (NTNs). However, NTNs face multiple routing and Quality of Service (QoS) provisioning challenges due to the mobility of network nodes. Distributed Software-Defined Networking (SDN) combined with Multi-Agent Deep Reinforcement Learning (MADRL) is widely used to introduce programmability and intelligent Traffic Engineering (TE) in TNs, yet applying DRL to NTNs is hindered by frequently changing state sizes, model scalability, and coordination issues. This paper introduces 3DQR, a novel TE framework that combines hierarchical multi-controller SDN, hierarchical MADRL based on Graph Neural Networks (GNNs), and network topology predictions for QoS path provisioning, effective load distribution, and flow rejection minimisation in future 3D networks. To enhance SDN scalability, introduced are metrics and path operations abstractions to facilitate domain agents coordination by the global agent. To the best of the authors’ knowledge, 3DQR is the first routing scheme to integrate MADRL and GNNs for optimising centralised routing and path allocation in SDN-based 3D mobile networks. The evaluations show up to a 14% reduction in flow rejection rate, a 50% improvement in traffic distribution, and effective QoS class prioritisation compared to baseline techniques. 3DQR also exhibits strong transfer capabilities, giving consistent performance gains in previously unseen environments. Full article

(This article belongs to the Special Issue Future Generation Non-Terrestrial Networks)

► Show Figures

Figure 1

25 pages, 2508 KiB

Open AccessArticle

OVSLT: Advancing Sign Language Translation with Open Vocabulary

by Ai Wang, Junhui Li, Wuyang Luan and Lei Pan

Electronics 2025, 14(5), 1044; https://doi.org/10.3390/electronics14051044 - 6 Mar 2025

Abstract

Hearing impairments affect approximately 1.5 billion individuals worldwide, highlighting the critical need for effective communication tools between deaf and hearing populations. Traditional sign language translation (SLT) models predominantly rely on gloss-based methods, which convert visual sign language inputs into intermediate gloss sequences before [...] Read more.

Hearing impairments affect approximately 1.5 billion individuals worldwide, highlighting the critical need for effective communication tools between deaf and hearing populations. Traditional sign language translation (SLT) models predominantly rely on gloss-based methods, which convert visual sign language inputs into intermediate gloss sequences before generating textual translations. However, these methods are constrained by their reliance on extensive annotated data, susceptibility to error propagation, and inadequate handling of low-frequency or unseen sign language vocabulary, thus limiting their scalability and practical application. Drawing upon multimodal translation theory, this study proposes the open-vocabulary sign language translation (OVSLT) method, designed to overcome these challenges by integrating open-vocabulary principles. OVSLT introduces two pivotal modules: Enhanced Caption Generation and Description (CGD), and Grid Feature Grouping with Advanced Alignment Techniques. The Enhanced CGD module employs a GPT model enhanced with a Negative Retriever and Semantic Retrieval-Augmented Features (SRAF) to produce semantically rich textual descriptions of sign gestures. In parallel, the Grid Feature Grouping module applies Grid Feature Grouping, contrastive learning, feature-discriminative contrastive loss, and balanced region loss scaling to refine visual feature representations, ensuring robust alignment with textual descriptions. We evaluated OVSLT on the PHOENIX-14T and CSLDaily datasets. The results demonstrated a ROUGE score of 29.6% on the PHOENIX-14T dataset and 30.72% on the CSLDaily dataset, significantly outperforming existing models. These findings underscore the versatility and effectiveness of OVSLT, showcasing the potential of open-vocabulary approaches to surpass the limitations of traditional SLT systems and contribute to the evolving field of multimodal translation. Full article

► Show Figures

Figure 1

23 pages, 8182 KiB

Open AccessArticle

Sound Source Localization Using Deep Learning for Human–Robot Interaction Under Intelligent Robot Environments

by Hong-Min Jo, Tae-Wan Kim and Keun-Chang Kwak

Electronics 2025, 14(5), 1043; https://doi.org/10.3390/electronics14051043 - 6 Mar 2025

Abstract

In this paper, we propose Sound Source Localization (SSL) using deep learning for Human–Robot Interaction (HRI) under intelligent robot environments. The proposed SSL method consists of three steps. The first step preprocesses the sound source to minimize noise and reverberation in the robotic [...] Read more.

In this paper, we propose Sound Source Localization (SSL) using deep learning for Human–Robot Interaction (HRI) under intelligent robot environments. The proposed SSL method consists of three steps. The first step preprocesses the sound source to minimize noise and reverberation in the robotic environment. Excitation source information (ESI), which contains only the original components of the sound source, is extracted from a sound source in a microphone array mounted on a robot to minimize background influence. Here, the linear prediction residual is used as the ESI. Subsequently, the cross-correlation signal between each adjacent microphone pair is calculated by using the ESI signal of each sound source. To minimize the influence of noise, a Generalized Cross-Correlation with the phase transform (GCC-PHAT) algorithm is used. In the second step, we design a single-channel, multi-input convolutional neural network that can independently learn the calculated cross-correlation signal between each adjacent microphone pair and the location of the sound source using the time difference of arrival. The third step classifies the location of the sound source after training with the proposed network. Previous studies have primarily used various features as inputs and stacked them into multiple channels, which made the algorithm complex. Furthermore, multi-channel inputs may not be sufficient to clearly train the interrelationship between each sound source. To address this issue, the cross-correlation signal between each sound source alone is used as the network input. The proposed method was verified on the Electronics and Telecommunications Research Institute-Sound Source Localization (ETRI-SSL) database acquired from the robotic environment. The experimental results revealed that the proposed method showed an 8.75% higher performance in comparison to the previous works. Full article

(This article belongs to the Special Issue Control and Design of Intelligent Robots)

► Show Figures

Figure 1

32 pages, 2960 KiB

Open AccessArticle

Comparing Application-Level Hardening Techniques for Neural Networks on GPUs

by Giuseppe Esposito, Juan-David Guerrero-Balaguera, Josie E. Rodriguez Condia and Matteo Sonza Reorda

Electronics 2025, 14(5), 1042; https://doi.org/10.3390/electronics14051042 - 6 Mar 2025

Abstract

Neural networks (NNs) are essential in advancing modern safety-critical systems. Lightweight NN architectures are deployed on resource-constrained devices using hardware accelerators like Graphics Processing Units (GPUs) for fast responses. However, the latest semiconductor technologies may be affected by physical faults that can jeopardize [...] Read more.

Neural networks (NNs) are essential in advancing modern safety-critical systems. Lightweight NN architectures are deployed on resource-constrained devices using hardware accelerators like Graphics Processing Units (GPUs) for fast responses. However, the latest semiconductor technologies may be affected by physical faults that can jeopardize the NN computations, making fault mitigation crucial for safety-critical domains. The recent studies propose software-based Hardening Techniques (HTs) to address these faults. However, the proposed fault countermeasures are evaluated through different hardware-agnostic error models neglecting the effort required for their implementation and different test benches. Comparing application-level HTs across different studies is challenging, leaving it unclear (i) their effectiveness against hardware-aware error models on any NN and (ii) which HTs provide the best trade-off between reliability enhancement and implementation cost. In this study, application-level HTs are evaluated homogeneously and independently by performing a study on the feasibility of implementation and a reliability assessment under two hardware-aware error models: (i) weight single bit-flips and (ii) neuron bit error rate. Our results indicate that not all HTs suit every NN architecture, and their effectiveness varies depending on the evaluated error model. Techniques based on the range restriction of activation function consistently outperform others, achieving up to 58.23% greater mitigation effectiveness while keeping the introduced overhead at inference time low while requiring a contained effort in their implementation. Full article

► Show Figures

Figure 1

12 pages, 4054 KiB

Open AccessArticle

Low-Frequency Communication Based on Rydberg-Atom Receiver

by Yipeng Xie, Mingwei Lei, Jianquan Zhang, Wenbo Dong and Meng Shi

Electronics 2025, 14(5), 1041; https://doi.org/10.3390/electronics14051041 - 6 Mar 2025

Abstract

Rydberg-atom receivers have developed rapidly with increasing sensitivity. However, studies on their application in low-frequency electric fields remain limited. In this work, we demonstrate low-frequency communication using an electrode-embedded atom cell and a whip antenna without the need for a low-noise amplifier (LNA). [...] Read more.

Rydberg-atom receivers have developed rapidly with increasing sensitivity. However, studies on their application in low-frequency electric fields remain limited. In this work, we demonstrate low-frequency communication using an electrode-embedded atom cell and a whip antenna without the need for a low-noise amplifier (LNA). Three modulations—binary phase-shift keying (BPSK), on–off keying (OOK), and two-frequency shift keying (2FSK)—were employed for communication using a Rydberg-atom receiver operating near 100 kHz. The signal-to-noise ratio (SNR) of the modulated low-frequency signal received by Rydberg atoms was measured at various emission voltages. Additionally, we demonstrated the in-phase and quadrature (IQ) constellation diagram, error vector magnitude (EVM), and eye diagram of the demodulated signal at different symbol rates. The EVM values were measured to be 8.8% at a symbol rate of 2 kbps, 9.4% at 4 kbps, and 13.7% at 8 kbps. The high-fidelity digital color image transmission achieved a peak signal-to-noise ratio (PSNR) of 70 dB. Our results demonstrate the feasibility of a Rydberg-atom receiver for low-frequency communication applications. Full article

(This article belongs to the Topic Quantum Wireless Sensing)

► Show Figures

Figure 1

Journal Description

Electronics

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI