Electronics

28 pages, 5745 KB

Open AccessEditor’s ChoiceArticle

FPGA-Based Design and Implementation of a High-Performance Telemetry Transmission Architecture for Satellite Communications

by Adriana N. Moreno Mercado and Víctor P. Gil Jiménez

Electronics 2026, 15(8), 1581; https://doi.org/10.3390/electronics15081581 - 10 Apr 2026

Viewed by 774

Abstract

This paper presents a high-performance and resource-efficient Field Programmable Gate Array (FPGA)-based architecture for satellite telemetry transmission systems. The proposed design implements a flexible channel coding chain, including Reed–Solomon (R-S) encoding, convolutional encoding, symbol interleaving, pseudo-randomization, and Attached Synchronization Marker (ASM) insertion, in [...] Read more.

This paper presents a high-performance and resource-efficient Field Programmable Gate Array (FPGA)-based architecture for satellite telemetry transmission systems. The proposed design implements a flexible channel coding chain, including Reed–Solomon (R-S) encoding, convolutional encoding, symbol interleaving, pseudo-randomization, and Attached Synchronization Marker (ASM) insertion, in accordance with CCSDS recommendations. The architecture is fully integrated and configurable, allowing dynamic selection of coding schemes without requiring structural modifications. The system is implemented on a modern FPGA platform with a 32-bit AXI4-Stream interface at 110 MHz, reaching an effective throughput of up to 1.76 Gbps. Experimental results demonstrate reliable timing with positive setup and hold margins, allowing the system to operate at approximately 130 MHz. Power consumption is measured using Switching Activity Interchange Format (SAIF)-based switching activity, providing a realistic estimate of programmable logic power consumption. The total on-chip power is about 1.77 W for individual coding modes. It rises to 1.91 W in the concatenated setup, which is the worst-case scenario. The results show that the proposed architecture efficiently uses resources, runs reliably at high speeds, and exhibits predictable power consumption. This makes it well suited for high-reliability and energy-constrained satellite communication systems. resources are used. Full article

(This article belongs to the Special Issue Advances in Satellite/UAV Communications)

► Show Figures

Figure 1

24 pages, 3589 KB

Open AccessEditor’s ChoiceArticle

Impact of Optimization Goal Visibility on Inter-Cloud DTM Performance

by Grzegorz Rzym, Zbigniew Duliński, Rafał Stankiewicz and Piotr Wydrych

Electronics 2026, 15(8), 1576; https://doi.org/10.3390/electronics15081576 - 9 Apr 2026

Viewed by 340

Abstract

This work presents an enhancement to the Dynamic Traffic Management (DTM) framework aimed at reducing signaling overhead between SDN controllers in multi-domain cloud environments. This extension is based on the ability to transmit information regarding the amount of balanced traffic and the optimal [...] Read more.

This work presents an enhancement to the Dynamic Traffic Management (DTM) framework aimed at reducing signaling overhead between SDN controllers in multi-domain cloud environments. This extension is based on the ability to transmit information regarding the amount of balanced traffic and the optimal transfer pattern. In the baseline periodic mode, the system regularly exchanges the compensation vector (

\vec{C}

) and the reference pattern (

\vec{R}

). To minimize communication, we define non-periodic modes that restrict

\vec{C}

updates and eliminate

\vec{R}

transmission entirely. Within these restricted signaling modes, we further distinguish between reactive and proactive operational schemes. Our experimental results demonstrate that reducing the visibility of optimization goals (

\vec{R}

and only sign of

\vec{C}

) and cutting signaling frequency in this manner maintains a comparable level of cost-efficiency. Specifically, the initial evaluation shows that DTM typically decreases transit costs by 8% to 15%, with maximum savings reaching up to 29% when compared to the worst-case default BGP path scenario. These findings suggest that the DTM mechanism can maintain its economic efficiency even with significantly reduced inter-domain coordination. Full article

(This article belongs to the Special Issue Recent Advances and Challenges in IoT, Cloud and Edge Coexistence, 2nd Edition)

► Show Figures

Figure 1

24 pages, 1821 KB

Open AccessEditor’s ChoiceArticle

MAVIS: Multi-Stem Audio Visualisation in Immersive Spaces Framework

by Jethro Shell and Sophy Smith

Electronics 2026, 15(8), 1559; https://doi.org/10.3390/electronics15081559 - 8 Apr 2026

Viewed by 668

Abstract

The visualisation of music has gained traction in both research and musical composition in recent years. The increased accessibility to immersive technologies, such as virtual reality (VR) and other forms of mixed reality (MR), lend themselves to the examination of how visualisation can [...] Read more.

The visualisation of music has gained traction in both research and musical composition in recent years. The increased accessibility to immersive technologies, such as virtual reality (VR) and other forms of mixed reality (MR), lend themselves to the examination of how visualisation can impact the perception of audio virtual worlds. In this paper, we propose the MAVIS (Multi-stem Audio Visualisation in Immersive Spaces) design framework, an approach to generating a visualisation of multi-stem structured orchestral music in a virtual world. This research explores the impact on participants’ interaction with an orchestral musical composition through the use of a two framework iterations informed by use cases. The resulting final design structure outlined in this article points towards constructing multi-stem virtual orchestral experiences through three pillars: semantic consistency, spatial agency, and complexity control. Whilst this research serves to propose a design intervention, future work requires a more extensive participant testing approach, coupled with an exploration of additional multimodal analysis. Full article

(This article belongs to the Special Issue Applications of Virtual, Augmented, and Mixed Reality in Creative Technology)

► Show Figures

Figure 1

25 pages, 11063 KB

Open AccessEditor’s ChoiceArticle

Tac-Mamba: A Pose-Guided Cross-Modal State Space Model with Trust-Aware Gating for mmWave Radar Human Activity Recognition

by Haiyi Wu, Kai Zhao, Wei Yao and Yong Xiong

Electronics 2026, 15(7), 1535; https://doi.org/10.3390/electronics15071535 - 7 Apr 2026

Viewed by 1010

Abstract

Millimeter-wave (mmWave) radar point clouds offer a privacy-preserving solution for Human Activity Recognition (HAR), but their inherent sparsity and noise limit single-modal performance. While multimodal fusion mitigates this issue, existing methods often suffer from severe negative transfer during visual degradation and incur high [...] Read more.

Millimeter-wave (mmWave) radar point clouds offer a privacy-preserving solution for Human Activity Recognition (HAR), but their inherent sparsity and noise limit single-modal performance. While multimodal fusion mitigates this issue, existing methods often suffer from severe negative transfer during visual degradation and incur high computational costs, unsuitable for edge devices. To address these challenges, we propose Tac-Mamba, a lightweight cross-modal state space model. First, we introduce a topology-guided distillation scheme that uses a Spatial Mamba teacher to extract structural priors from visual skeletons. These priors are then explicitly distilled into a Point Transformer v3 (PTv3) radar student with a modality dropout strategy. We also developed a Trust-Aware Cross-Modal Attention (TACMA) module to prevent negative transfer. It evaluates the reliability of visual features through a SiLU-activated cross-modal bilinear interaction, smoothly degrading to a pure radar-driven fallback projection when visual inputs are corrupted. Finally, a Lightweight Temporal Mamba Block (LTMB) with a Zero-Parameter Cross-Gating (ZPCG) mechanism captures long-range kinematic dependencies with linear complexity. Experiments on the public MM-Fi dataset under strict cross-environment protocols demonstrate that Tac-Mamba achieves competitive accuracies of 95.37% (multimodal) and 87.54% (radar-only) with only 0.86M parameters and 1.89 ms inference latency. These results highlight the model’s exceptional robustness to modality missingness and its feasibility for edge deployment. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Graphical abstract

26 pages, 2454 KB

Open AccessEditor’s ChoiceArticle

Evolving LLMs from Next-Token Prediction to Multi-Token Prediction via Self-Distillation

by Yang Xu and Wanxiang Che

Electronics 2026, 15(7), 1533; https://doi.org/10.3390/electronics15071533 - 6 Apr 2026

Viewed by 2628

Abstract

Mainstream Large Language Models (LLMs) work under the paradigm of Next-Token Prediction (NTP). Multi-Token Prediction (MTP) is motivated by higher decoding efficiency, extending NTP to enable LLMs to draft multiple tokens during each forward pass. However, existing MTP approaches pretrain MTP along with [...] Read more.

Mainstream Large Language Models (LLMs) work under the paradigm of Next-Token Prediction (NTP). Multi-Token Prediction (MTP) is motivated by higher decoding efficiency, extending NTP to enable LLMs to draft multiple tokens during each forward pass. However, existing MTP approaches pretrain MTP along with the target LLM, making it difficult to unlock MTP for LLMs without official support. In this work, we propose a post-hoc approach to training an MTP module for a target LLM, providing an efficient way to evolve the LLM from NTP to MTP. The proposed approach features two main characteristics. (1) No changes to the target LLM, since it is frozen during MTP training. (2) Efficient MTP training via self-distillation from the target LLM’s native NTP capability. Results show that our approach can post-hoc train a performant MTP module via lightweight pretraining. Full article

(This article belongs to the Special Issue Innovative Applications of Large Language Models in Natural Language Processing (NLP))

► Show Figures

Figure 1

23 pages, 2145 KB

Open AccessEditor’s ChoiceArticle

Seeing Through Touch: A Stereo-Vision Vibrotactile Aid for Visually Impaired People

by Claudia Presicci, Giulia Ballardini, Giorgia Marchesi, Paolo Robutti, Matteo Moro, Camilla Pierella, Andrea Canessa and Maura Casadio

Electronics 2026, 15(7), 1511; https://doi.org/10.3390/electronics15071511 - 3 Apr 2026

Viewed by 720

Abstract

Blind and visually impaired individuals face persistent challenges when navigating unfamiliar environments, where unseen obstacles compromise their safety and independence. Although many electronic travel aids have been proposed, most remain impractical for daily use—they often rely on bulky or costly hardware, require external [...] Read more.

Blind and visually impaired individuals face persistent challenges when navigating unfamiliar environments, where unseen obstacles compromise their safety and independence. Although many electronic travel aids have been proposed, most remain impractical for daily use—they often rely on bulky or costly hardware, require external processing, or provide unintuitive feedback. This work presents a wearable stereo-vision-based vibrotactile system for real-time obstacle detection and navigation assistance. The device combines an off-the-shelf stereo camera integrated with a simultaneous localization and mapping framework to perceive spatial geometry and detect obstacles in the user’s path. Two stereo-matching methods were implemented to estimate depth: a block-based algorithm optimized for low-latency performance and a semi-global approach providing denser depth maps. Detected obstacles are translated into distinct vibration patterns delivered through four skin-contact body-mounted actuators encoding both direction and distance. The system was evaluated with blindfolded sighted, visually impaired, and blind participants. Both stereo approaches supported reliable real-time guidance and high obstacle-avoidance rates, demonstrating robust performance on affordable, wearable hardware. These findings confirm the feasibility of real-time tactile guidance using commercially available components, marking a concrete step toward accessible navigation support that enhances safety and autonomy for blind and visually impaired individuals. Full article

(This article belongs to the Special Issue Feature Papers in Bioelectronics: 2025–2026 Edition)

► Show Figures

Figure 1

19 pages, 1674 KB

Open AccessEditor’s ChoiceArticle

Phaseless Characterization of Multilayered Media: Combining Interferometric Holography and a MUSIC-Based Approach

by Mario Del Prete, Raffaele Solimene, Loreto Di Donato and Maria Antonia Maisto

Electronics 2026, 15(7), 1496; https://doi.org/10.3390/electronics15071496 - 2 Apr 2026

Viewed by 504

Abstract

Millimeter-wave and sub-millimeter-wave techniques are widely used in non-destructive testing of multilayered materials due to their ability to penetrate non-conductive media and resolve dielectric stratifications. However, conventional thickness estimation methods suffer from an inherent resolution limit dictated by the available frequency bandwidth. In [...] Read more.

Millimeter-wave and sub-millimeter-wave techniques are widely used in non-destructive testing of multilayered materials due to their ability to penetrate non-conductive media and resolve dielectric stratifications. However, conventional thickness estimation methods suffer from an inherent resolution limit dictated by the available frequency bandwidth. In this paper, a MUSIC-based approach is proposed to achieve super-resolution localization of echoes in the reflective response of the structure under test. The method exploits the sparsity of the reflective response, similarly to compressive sensing approaches, while providing improved reconstruction accuracy. Moreover, the proposed strategy enables the retrieval of dielectric permittivities and layer thicknesses without resorting to complex nonlinear fitting procedures. Finally, the method operates on magnitude-only data, with phase information recovered through an interferometric holographic technique, making the proposed framework well-suited for cost-effective industrial applications. Full article

(This article belongs to the Special Issue Inverse Problems and Optimization in Electromagnetic Systems)

► Show Figures

Figure 1

33 pages, 8263 KB

Open AccessEditor’s ChoiceArticle

Semantic Graphs of Learning Activities from LLM Embeddings: A Lightweight and Explainable Approach for Smart Learning Systems

by Javier García-Sigüenza, Alberto Real-Fernández, Faraón Llorens-Largo, Jose F. Vicent and Rafael Molina-Carmona

Electronics 2026, 15(7), 1414; https://doi.org/10.3390/electronics15071414 - 28 Mar 2026

Viewed by 729

Abstract

Smart learning systems are designed to analyze the context, needs, and progress of each student. These are becoming increasingly common, but they present challenges, such as predicting student performance and automatically managing learning activities. In this context, Large Language Models (LLMs) can be [...] Read more.

Smart learning systems are designed to analyze the context, needs, and progress of each student. These are becoming increasingly common, but they present challenges, such as predicting student performance and automatically managing learning activities. In this context, Large Language Models (LLMs) can be useful, as they are capable of understanding word relationships and analyzing their context. They are often associated with chatbots, which are computationally expensive, thereby complicating their integration. Instead, in this work, we propose to leverage the capabilities of LLMs through a semantic graph of activities created from sentence embeddings. This representation is a lightweight and explainable alternative. On the one hand, it requires a lower computational cost. On the other hand, it allows us to observe which activities are most similar directly. On this basis, we propose two problems to validate our proposal. In the first, we use the graph to classify new activities. In the second, we extend this representation with the temporal dimension to formulate a spatio-temporal problem and predict student performance. The results show that the semantic graph not only provides an accurate representation for the organization and classification of activities, but also offers practical advantages and improves explainability. Full article

(This article belongs to the Special Issue New Insights into Natural Language Processing and Large Language Models)

► Show Figures

Figure 1

13 pages, 44672 KB

Open AccessEditor’s ChoiceArticle

ARMANI: Dictionary-Learning-Inspired Data-Free Deep Generative Modeling with Meta-Attention and Implicit Preconditioning for Compressively Sampled Magnetic Resonance Imaging

by Ming Wu, Jing Cheng, Qingyong Zhu and Dong Liang

Electronics 2026, 15(7), 1402; https://doi.org/10.3390/electronics15071402 - 27 Mar 2026

Viewed by 491

Abstract

Magnetic resonance imaging (MRI) reconstruction from undersampled k-space data enables accelerated acquisition but leads to a severely ill-posed inverse problem. Although supervised deep learning methods have achieved strong performance, they typically rely on large paired datasets that are difficult to obtain in clinical [...] Read more.

Magnetic resonance imaging (MRI) reconstruction from undersampled k-space data enables accelerated acquisition but leads to a severely ill-posed inverse problem. Although supervised deep learning methods have achieved strong performance, they typically rely on large paired datasets that are difficult to obtain in clinical practice. To address these limitations, we propose a dictionary-learning-inspired dAta-fRee deep generative modeling with Meta-Attention and implicit precoNditIoning for compressively sampled MRI (CS-MRI), termed ARMANI. Specifically, a meta-attention-augmented deep image prior (MA-DIP) generator performs a joint optimization over the latent input

η

and the network parameter

θ

, where

η

is regularized via gradient-domain sparsity and

θ

is constrained by a ridge penalty, mirroring the adaptive estimation of sparse coefficients and an empirical sparsifying dictionary. Furthermore, we integrate a single-step pseudo-orthogonal projection to achieve implicit preconditioning, which modulates the loss landscape and mitigates ill-conditioning of the forward operator. Experimental results demonstrate that ARMANI consistently outperforms existing SOTA data-free and self-supervised methods, and, with limited training data, achieves performance comparable to or slightly better than the supervised benchmark MoDL, with effective artifact suppression and faithful recovery of fine structural details. Overall, ARMANI shows strong scalability and potential for practical deployment in fully data-free CS-MRI reconstruction scenarios. Full article

(This article belongs to the Special Issue Artificial Intelligence and Machine Learning for Medical Image Processing)

► Show Figures

Figure 1

42 pages, 2464 KB

Open AccessEditor’s ChoiceArticle

Energy-Aware Multilingual Evaluation of Large Language Models

by I. de Zarzà, Mauro Liz, J. de Curtò and Carlos T. Calafate

Electronics 2026, 15(7), 1395; https://doi.org/10.3390/electronics15071395 - 27 Mar 2026

Cited by 3 | Viewed by 1122

Abstract

The rapid deployment of Large Language Models (LLMs) in multilingual, production-scale systems has made inference-time energy consumption a critical yet systematically under-evaluated dimension of model quality. While accuracy-centric benchmarks dominate current evaluation practice, they fail to capture the energy cost of reasoning, particularly [...] Read more.

The rapid deployment of Large Language Models (LLMs) in multilingual, production-scale systems has made inference-time energy consumption a critical yet systematically under-evaluated dimension of model quality. While accuracy-centric benchmarks dominate current evaluation practice, they fail to capture the energy cost of reasoning, particularly across languages and task complexities where consumption profiles diverge substantially. In this work, we present a comprehensive energy–performance evaluation of five instruction-tuned LLMs, spanning Transformer, Grouped-Query Attention, and State Space Model architectures, across thirteen typologically diverse languages and multiple task difficulty levels under controlled GPU-level energy measurement on NVIDIA H200 hardware. Our analysis encompasses 65 model–language configurations totaling over 5100 individual inference runs, supported by rigorous non-parametric statistical testing (Friedman tests, pairwise Wilcoxon signed-rank with Holm correction, and paired Cohen’s d effect sizes). We report four principal findings. First, energy consumption varies up to threefold across models under identical workloads (

χ^{2} = 49.42

,

p = 4.78 \times 10^{- 10}

, Friedman test), stratifying into three distinct energy regimes driven by architecture and generation dynamics rather than parameter count. Second, energy expenditure and reasoning performance are only weakly coupled, as confirmed by Spearman rank correlation analysis (

r_{s} = 0.109

,

p = 0.386

). Third, task category and difficulty level introduce substantial and model-dependent variation in both energy demand and performance, with cross-lingual performance variance amplifying at higher difficulty levels. Fourth, language choice acts as a measurable deployment parameter as follows: Romance languages on average achieve lower energy consumption than English across multiple models, while model efficiency rankings shift across languages, yielding language-dependent Pareto-optimal frontiers. We formalize these trade-offs through multi-objective Pareto analysis and introduce a composite AI Energy Score metric that captures reasoning quality per unit of energy. Of the 65 evaluated configurations, only four are Pareto-optimal, three Mistral-7B configurations at the low-energy extreme and one Phi-4-mini-instruct configuration at the high-performance end, while three of the five models are entirely dominated across all language configurations. These findings provide actionable guidelines for energy-aware model selection in multilingual deployments and support the integration of AI Energy Scores as a standard complementary criterion in LLM evaluation frameworks. Full article

(This article belongs to the Special Issue Data-Related Challenges in Machine Learning: Theory and Application)

► Show Figures

Figure 1

15 pages, 1836 KB

Open AccessEditor’s ChoiceArticle

Numerical Simulation and Optimization of Dark Current Performance Through a Quaternary Barrier in InAs/GaSb Superlattice Photodetectors

by Zhejing Jiao, Gaoyu Zhou, Xin Jin, Yi Gu, Bowen Liu, Tao Li and Xue Li

Electronics 2026, 15(7), 1355; https://doi.org/10.3390/electronics15071355 - 25 Mar 2026

Viewed by 725

Abstract

In this work, a high-performance mid-wave infrared (MWIR) photodetector (PD) utilizing an InAs/GaSb Type-II superlattice absorber and a quaternary AlGaAsSb barrier is designed and analyzed based on numerical simulations aimed at determining an optimized detector structure. Through these simulations, the composition of the [...] Read more.

In this work, a high-performance mid-wave infrared (MWIR) photodetector (PD) utilizing an InAs/GaSb Type-II superlattice absorber and a quaternary AlGaAsSb barrier is designed and analyzed based on numerical simulations aimed at determining an optimized detector structure. Through these simulations, the composition of the AlGaAsSb barrier is carefully designed to achieve lattice matching, high conduction band offset and zero valence band offset. By optimizing the barrier thickness and doping concentration, the depletion region is effectively shifted from the narrow-bandgap absorber to the wide-bandgap barrier; additionally, at 150 K and a reversed bias of 0.05 V, the dark current density in the PD with the barrier (pBn) is reduced to 1.83 × 10⁻⁵ A/cm², about two orders of magnitude lower than that of the PD without the barrier. Furthermore, the effect of the barrier on the generation–recombination (G-R) and the trap-assisted tunneling (TAT) currents are analyzed and compared in detail, and it is found that the barrier structure is much more effective in suppressing the TAT current at low reversed bias when the electric field is low in the absorber layer. These results demonstrate the efficacy of the proposed AlGaAsSb barrier design for realizing high-operating-temperature MWIR PDs. It also provides an insight into the physical mechanism that leads to the performance enhancement of InAs/GaSb PDs. Full article

(This article belongs to the Special Issue Feature Papers in Semiconductor Devices, 2nd Edition)

► Show Figures

Figure 1

26 pages, 16104 KB

Open AccessEditor’s ChoiceArticle

Multi-Slot Attention with State Guidance for Egocentric Robotic Manipulation

by Sofanit Wubeshet Beyene and Ji-Hyeong Han

Electronics 2026, 15(7), 1365; https://doi.org/10.3390/electronics15071365 - 25 Mar 2026

Viewed by 857

Abstract

Visual perception is fundamental to robotic manipulation for recognizing objects, goals, and contextual details. Third-person cameras provide global views but can miss contact-rich interactions and require calibration. Wrist-mounted egocentric cameras reduce these limitations but introduce occlusion, motion blur, and partial observability, which complicate [...] Read more.

Visual perception is fundamental to robotic manipulation for recognizing objects, goals, and contextual details. Third-person cameras provide global views but can miss contact-rich interactions and require calibration. Wrist-mounted egocentric cameras reduce these limitations but introduce occlusion, motion blur, and partial observability, which complicate visuomotor learning. Furthermore, existing perception modules that rely solely on pixels or fuse imagery with proprioception as flat vectors do not explicitly model structured scene representations in dynamic egocentric views. To address these challenges, a multi-slot attention fusion encoder for egocentric manipulation is introduced. Learnable slot queries extract localized visual features from image tokens, and Feature-wise Linear Modulation (FiLM) conditions each slot on the robot’s joint states, producing a structured slot-based latent representation that adapts to viewpoint and configuration changes without requiring object labels or external camera priors. The resulting structured slot-based latent representation is used as input to a Soft Actor–Critic (SAC) agent, which achieves a higher mean cumulative return than pixel-only CNN/DrQ and state-only baselines on a ManiSkill3 egocentric manipulation task. Probing experiments and real-camera evaluation further show that the learned representation remains stable under egocentric viewpoint shifts and partial occlusions, indicating robustness in practical manipulation settings. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

19 pages, 635 KB

Open AccessEditor’s ChoiceArticle

Conformal Prediction for Counterfactual Detection in Concept Learning from Synthetic Visual Patterns

by Ulf Norinder, Stephanie Lowry, Heimo Müller and Andreas Holzinger

Electronics 2026, 15(7), 1346; https://doi.org/10.3390/electronics15071346 - 24 Mar 2026

Cited by 1 | Viewed by 795

Abstract

Reliable detection of previously unseen classes under distributional shift remains a central challenge in concept learning and explainable artificial intelligence. In particular, high-performance deep learning models often lack statistically grounded mechanisms to signal when an instance deviates from learned concepts. This paper addresses [...] Read more.

Reliable detection of previously unseen classes under distributional shift remains a central challenge in concept learning and explainable artificial intelligence. In particular, high-performance deep learning models often lack statistically grounded mechanisms to signal when an instance deviates from learned concepts. This paper addresses this limitation by investigating whether conformal prediction can be effectively combined with a YOLOv5 deep learning classifier to enable principled counterfactual detection without prior exposure to the counterfactual class. As a controlled testbed, we employ Kandinsky patterns, a structured benchmark widely used in explainable AI research due to its rule-based generative transparency and suitability for concept learning studies. The proposed framework first classifies valid and invalid patterns and subsequently applies inductive conformal prediction to obtain calibrated prediction sets at a user-defined significance level. Counterfactual instances are, at start, identified based solely on information from known true and false patterns, without explicit training examples of the counterfactual class. Experimental results demonstrate that the conformalized detector reliably identifies a substantial proportion of previously unseen counterfactual patterns while maintaining statistical validity. In addition, the method flags unlabeled (“empty”) instances, thereby providing a principled signal for the emergence of new concepts. By conformalizing YOLOv5 outputs, the approach establishes a statistically sound mechanism for uncertainty-aware detection of divergent classes, contributing to robust and explainable concept learning in structured visual pattern recognition. Full article

(This article belongs to the Special Issue Artificial Intelligence for Smart Image Perception, Recognition and Understanding)

► Show Figures

Figure 1

10 pages, 2043 KB

Open AccessEditor’s ChoiceArticle

Vortex Wave Generation at E-Band Using a TWT Source and Metasurface

by Haojie Zhu, Jinjun Feng, Pan Pan, Shishuo Liu, Yueyi Zhang and Chaohai Du

Electronics 2026, 15(7), 1348; https://doi.org/10.3390/electronics15071348 - 24 Mar 2026

Viewed by 486

Abstract

In this paper, a novel scheme is introduced that combines a traveling wave tube (TWT) with a metasurface to generate high-power E-band vortex electromagnetic waves. The TE₁₀ mode electromagnetic wave emitted by the TWT is initially converted into a plane wave via [...] Read more.

In this paper, a novel scheme is introduced that combines a traveling wave tube (TWT) with a metasurface to generate high-power E-band vortex electromagnetic waves. The TE₁₀ mode electromagnetic wave emitted by the TWT is initially converted into a plane wave via a horn antenna and subsequently transformed into a vortex electromagnetic wave by the metasurface. The metasurface is designed and simulated, and the results show that this approach can convert the TE₁₀ mode from the TWT into vortex electromagnetic waves with a specific topological charge of

l = + 1

within the 71–76 GHz frequency range, achieving a remarkable mode purity of up to 97%. The experiment at 73.5 GHz was successfully carried out, generating vortex electromagnetic waves with the designated topological charge of

l = + 1

using this method. Although the experimentally measured mode purity was limited to 30.6%, this outcome confirms the effectiveness of the proposed method. Full article

(This article belongs to the Special Issue Advanced RF, Microwave, and Millimeter-Wave Antennas, Devices, and Systems)

► Show Figures

Graphical abstract

26 pages, 2609 KB

Open AccessEditor’s ChoiceArticle

Scale, Structure, and Stability: When Does LLM-Based Data Augmentation Improve Temporal Robustness in Web Intrusion Detection?

by Jun yeop Lee and Hee Cheol Kim

Electronics 2026, 15(7), 1344; https://doi.org/10.3390/electronics15071344 - 24 Mar 2026

Viewed by 773

Abstract

We investigate when LLM-based data augmentation can mitigate temporal collapse in web intrusion detection under extreme cross-temporal distribution shift. Under a strict hold-out protocol—training on CSIC-2010 and evaluating exclusively on the temporally separated SRBH-2020 golden set—the legacy-trained baseline exhibits near-collapse in balanced correlation [...] Read more.

We investigate when LLM-based data augmentation can mitigate temporal collapse in web intrusion detection under extreme cross-temporal distribution shift. Under a strict hold-out protocol—training on CSIC-2010 and evaluating exclusively on the temporally separated SRBH-2020 golden set—the legacy-trained baseline exhibits near-collapse in balanced correlation (MCC ≈ 0) despite retaining high recall, revealing severe false-positive bias under drift. Rather than assuming uniform benefits from synthetic data, we analyze how augmentation effects vary with model scale. Using a fixed DeBERTa-v3-base backbone and five random seeds, we compare synthetic training corpora generated by multiple LLMs under identical schema-guided structured decoding and filtering constraints. The results reveal a scale-dependent threshold effect. Models below 12B parameters (i.e., the 4–8B settings in our experiments) frequently introduce structural artifacts that amplify false-positive bias and further destabilize cross-temporal performance. In contrast, models at or above 12B parameters consistently produce modest but statistically reliable recovery from correlation collapse (p < 0.001), with balanced metrics shifting toward the target-domain distribution. Although the absolute performance remains limited under extreme temporal separation, a confusion-matrix analysis shows that large-scale generation reduces false-positive skew and moves decision-boundary behavior closer to the modern-domain regime. These findings indicate that LLM-based augmentation is not inherently robustness-enhancing; rather, its effect depends critically on model scale and disciplined generation control. When properly constrained, ≥12B-scale models can partially stabilize cross-temporal behavior, whereas smaller-scale generation may exacerbate distributional fragility. Full article

(This article belongs to the Special Issue AI and Cybersecurity: Emerging Trends and Key Challenges)

► Show Figures

Figure 1

31 pages, 4949 KB

Open AccessEditor’s ChoiceArticle

Attention Distribution-Aware Softmax for NPU-Accelerated On-Device Inference of LLMs: An Edge-Oriented Approximation Design

by Sanoop Sadheerthan, Min-Jie Hsu, Chih-Hsiang Huang and Yin-Tien Wang

Electronics 2026, 15(6), 1312; https://doi.org/10.3390/electronics15061312 - 20 Mar 2026

Cited by 1 | Viewed by 1444

Abstract

Low-power NPUs enable on-device LLM inference through efficient integer and fixed-point algebra, yet their lack of native exponential support makes Transformer softmax a critical performance bottleneck. Existing NPU kernels approximate

e^{x}

using uniform piecewise polynomials to enable O(1) SIMD indexing, but this [...] Read more.

Low-power NPUs enable on-device LLM inference through efficient integer and fixed-point algebra, yet their lack of native exponential support makes Transformer softmax a critical performance bottleneck. Existing NPU kernels approximate

e^{x}

using uniform piecewise polynomials to enable O(1) SIMD indexing, but this wastes computation by applying high-degree arithmetic indiscriminately in every segment. Conversely, fully adaptive approaches maximize statistical fidelity but introduce pipeline stalls due to comparator-based boundary search. To bridge this gap, we propose an attention distribution-aware softmax that uses Particle Swarm Optimization (PSO) to define non-uniform segments and variable polynomial degrees, prioritizing finer granularity and lower arithmetic complexity in attention-dense regions. To ensure efficiency, we snap boundaries into a 128-bin LUT, enabling O(1) retrieval of segment parameters without branching. Inference measurements show that this favors low-degree execution, minimizing exp-kernel overhead. Using TinyLlama-1.1B-Chat as a testbed, the proposed weighted design reduces cycles per call exp kernel (CPC) by 18.5% versus an equidistant uniform Degree-4 baseline and 13.1% versus uniform Degree-3, while preserving ranking fidelity. These results show that grid-snapped, variable-degree approximation can improve softmax efficiency while largely preserving attention ranking fidelity, enabling accurate edge LLM inference. Full article

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

► Show Figures

Figure 1

16 pages, 6917 KB

Open AccessEditor’s ChoiceArticle

Design of a Receiver Path with Self-Developed Limiter MMIC of X-Band for AESA Radar Systems

by Yuseok Jeon, Jaejin Koo, Minseok Ahn and Youngoo Yang

Electronics 2026, 15(6), 1272; https://doi.org/10.3390/electronics15061272 - 18 Mar 2026

Viewed by 1481

Abstract

In the present study, the limiter component with excellent low insertion loss and leakage power characteristics is used at the beginning of the receiving path, mounted at the rear end of the antenna of the AESA radar system, to protect the low noise [...] Read more.

In the present study, the limiter component with excellent low insertion loss and leakage power characteristics is used at the beginning of the receiving path, mounted at the rear end of the antenna of the AESA radar system, to protect the low noise amplifier (LNA) from excessive input power. The main components required for the X-band transmit/receive module are designed and manufactured mainly using bare-type components to reduce the module size. In this paper, we develop the limiter component, which is a key component, and verify whether it can secure performance that can be operated from the system perspective by mounting it on the receiving path of the transmit/receive module. The performance results of the limiter component unit obtained insertion loss of less than 0.615 dB at 10 GHz and leakage power of less than +16.8 dBm in the X-band. The main performance of the receiving path in the transmit/receive module unit obtained results of a noise figure of less than 3.2 dB and a gain of more than 37 dB (including two stages of LNA). Full article

(This article belongs to the Section Microwave and Wireless Communications)

► Show Figures

Figure 1

38 pages, 3950 KB

Open AccessEditor’s ChoiceArticle

Investigating Post-Quantum Cryptography to Secure Transmitted Data via Mobile Communication

by Rongjie Zhou, Huaqun Guo and Francis Ee Cheok Teo

Electronics 2026, 15(6), 1275; https://doi.org/10.3390/electronics15061275 - 18 Mar 2026

Cited by 1 | Viewed by 1651

Abstract

The advent of quantum computing poses significant challenges to traditional cryptographic systems, threatening the confidentiality, integrity and authenticity of digital communications. This paper investigates the integration of post-quantum cryptography (PQC) algorithms into mobile communication systems to address these challenges. The study focuses on [...] Read more.

The advent of quantum computing poses significant challenges to traditional cryptographic systems, threatening the confidentiality, integrity and authenticity of digital communications. This paper investigates the integration of post-quantum cryptography (PQC) algorithms into mobile communication systems to address these challenges. The study focuses on evaluating key PQC algorithms shortlisted by the National Institute of Standards and Technology (NIST), including CRYSTALS-Kyber, CRYSTALS-Dilithium, Falcon and SPHINCS+, within the context of 5G and future mobile network architectures. The research encompasses the design and implementation of an experimental framework involving mobile devices, servers, and cloud-based infrastructure to simulate real-world communication scenarios. Performance metrics such as key generation time, signature generation, encryption and decryption speed, and resource consumption were analyzed across various devices to identify algorithms suitable for mobile environments. The findings reveal that lattice-based algorithms, such as Kyber and Dilithium, offer a promising balance between security and efficiency, making them ideal for resource-constrained devices. In contrast, hash-based algorithms like SPHINCS+ exhibit higher computational demands, limiting their practicality in certain applications. This work highlights the importance of algorithm selection and hardware optimization in ensuring secure and efficient communications in the quantum era. By integrating theoretical advancements in PQC with practical applications, this research lays the foundation for quantum-resistant security in mobile networks, ensuring secure and future-ready digital communications. Full article

(This article belongs to the Special Issue New Technologies for Cybersecurity)

► Show Figures

Figure 1

27 pages, 3484 KB

Open AccessEditor’s ChoiceArticle

Enhancing RMF and ATT&CK Mapping Accuracy Through Integration of Sentence-BERT and Mitigation Parameters

by Hanhee Lee, Sukjoon Yoon, Yunkyung Lee and Jiwon Kang

Electronics 2026, 15(6), 1248; https://doi.org/10.3390/electronics15061248 - 17 Mar 2026

Viewed by 761

Abstract

To minimize cybersecurity risks in weapon systems, the implementation of the Korean Risk Management Framework (K-RMF) has become imperative. However, a significant “strategic gap” exists between high-level RMF controls and technical MITRE ATT&CK techniques, rendering manual mapping labor-intensive. This study proposes an automated [...] Read more.

To minimize cybersecurity risks in weapon systems, the implementation of the Korean Risk Management Framework (K-RMF) has become imperative. However, a significant “strategic gap” exists between high-level RMF controls and technical MITRE ATT&CK techniques, rendering manual mapping labor-intensive. This study proposes an automated mitigation-driven pipeline that integrates Sentence-BERT (SBERT) with the structural defense relationships of the ATT&CK knowledge graph. To address the data coverage limitations of the Center for Threat-Informed Defense (CTID) silver standard, we introduce Recall@restricted as a calibrated performance metric. Experimental evaluations demonstrate that the proposed ensemble framework achieves a Recall@restricted of 0.74, significantly outperforming baseline SBERT-only models. These findings suggest that deterministic mitigation relationships effectively complement semantic representations, providing a robust framework for aligning RMF controls with adversarial behaviors. Full article

(This article belongs to the Special Issue Emerging Research Trends and Technologies in Intrusion Detection Systems (IDSs) and Artificial Intelligence (AI) Utilization)

► Show Figures

Figure 1

21 pages, 656 KB

Open AccessEditor’s ChoiceArticle

Acoustic Violence Detection Using Cascade Strategy for Computationally Constrained Scenarios

by Fangfang Zhu-Zhou, Diana Tejera-Berengué, Roberto Gil-Pita, Manuel Utrilla-Manso and Manuel Rosa-Zurera

Electronics 2026, 15(6), 1227; https://doi.org/10.3390/electronics15061227 - 16 Mar 2026

Viewed by 1142

Abstract

Detecting violent content in audio recordings is crucial for public safety, autonomous surveillance, and content moderation, particularly when visual cues are unreliable or unavailable. A resource-aware two-stage cascade system is proposed for acoustic violence detection that combines a lightweight Least Squares Linear Detector [...] Read more.

Detecting violent content in audio recordings is crucial for public safety, autonomous surveillance, and content moderation, particularly when visual cues are unreliable or unavailable. A resource-aware two-stage cascade system is proposed for acoustic violence detection that combines a lightweight Least Squares Linear Detector (LSLD) as a first-stage screener with a trimmed version of YAMNet as a second-stage classifier. A percentile-based forwarding rule controls the fraction of segments routed to the deep stage, turning the accuracy–cost trade-off into an explicit operating parameter for always-on deployment. The approach is evaluated on a publicly released dataset of real-world violent audio augmented with background noise and artificial reverberation. The results in the low-false-alarm regime show that the proposed cascade preserves performance close to a Stage 2-only baseline while substantially reducing average deep-inference workload. An ablation study validates the role of the LSLD as an inexpensive pre-filter, and robustness is assessed under clean, reverberant, and 12 dB noise conditions. Finally, an analytic energy consumption model is provided, which links computational workload to daily energy demand and photovoltaic sizing on ultra-low-power hardware, supporting sustainable off-grid deployment. Full article

(This article belongs to the Special Issue Implementation of Neural Network Models on Resource-Constrained Devices)

► Show Figures

Figure 1

24 pages, 10576 KB

Open AccessEditor’s ChoiceArticle

Accurate Road User Position Estimation for V2I Using Point Clouds from Mobile Mapping Systems

by Ju Hee Yoo, Ho Gi Jung and Jae Kyu Suhr

Electronics 2026, 15(6), 1238; https://doi.org/10.3390/electronics15061238 - 16 Mar 2026

Viewed by 451

Abstract

Accurate detection and positioning of road users are essential for vehicle-to-infrastructure (V2I)-assisted autonomous driving. For this purpose, the road user’s ground contact point is usually detected in a monocular camera image. Then, a homography-based method is used to convert this detected point into [...] Read more.

Accurate detection and positioning of road users are essential for vehicle-to-infrastructure (V2I)-assisted autonomous driving. For this purpose, the road user’s ground contact point is usually detected in a monocular camera image. Then, a homography-based method is used to convert this detected point into its corresponding map position. However, the homography-based method assumes that the ground is planar, which leads to significant positioning errors in real-world environments. This limitation degrades the reliability of V2I-assisted autonomous driving, particularly in environments with complex road geometries. This study presents a method for accurately estimating the positions of road users using 3D point clouds generated by a Mobile Mapping System (MMS) for map construction without incurring additional costs. Moreover, since surveillance cameras are typically installed in urban areas, point clouds for these regions are often already available. The proposed method uses a pre-generated Look-Up Table (LUT), which is created by projecting MMS-based 3D point clouds onto the image coordinate system, so that each pixel in the image stores its corresponding 3D map position. Once the ground contact points of road users are detected in the image, the corresponding 3D positions on the map can be directly obtained by referencing the LUT. In the experiments, the proposed method was evaluated using surveillance camera images and MMS-based point clouds collected from various real-world environments. The results show that the proposed method reduces positioning errors of road users by an average of 61.4% compared to the conventional homography-based method. The improvement is particularly significant in environments with ground slope variations. In addition, the proposed method demonstrates real-time feasibility on an embedded camera, achieving low latency and power-efficient performance suitable for V2I edge deployment. Full article

(This article belongs to the Special Issue Autonomous Vehicles: Sensing, Mapping, and Positioning)

► Show Figures

Figure 1

24 pages, 905 KB

Open AccessEditor’s ChoiceArticle

Neural Encoding Strategies for Neuromorphic Computing

by Michael Liu, Honghao Zheng and Yang Yi

Electronics 2026, 15(6), 1221; https://doi.org/10.3390/electronics15061221 - 14 Mar 2026

Viewed by 1248

Abstract

Neuromorphic computing seeks to mimic structure and function of biological neural systems to enable energy-efficient, adaptive information processing. A critical component of this paradigm is neural encoding—the translation of analog or digital input data into spike-based representations suitable for spiking neural networks (SNNs). [...] Read more.

Neuromorphic computing seeks to mimic structure and function of biological neural systems to enable energy-efficient, adaptive information processing. A critical component of this paradigm is neural encoding—the translation of analog or digital input data into spike-based representations suitable for spiking neural networks (SNNs). This paper provides a comprehensive overview of major neural encoding schemes used in neuromorphic systems, including rate and temporal encoding, as well as latency, interspike interval, phase, and multiplexed encoding. The purpose of this paper is to explore the use of encoding techniques for deep learning applications. We discussed the underlying principles of spike encoding approaches, their biological inspiration, computational efficiency, power consumption, integrated circuit design and implementation, and suitability for various neuromorphic applications. We also presented our research on a hardware-and-software co-design platform for different encoding schemes and demonstrated their performance. By comparing their strengths, limitations, and implementation challenges, we aim to provide insights that will guide the development of more efficient and application-specific neuromorphic systems. We also performed an encoder performance analysis via Python 3.12 simulations to compare classification accuracies across these spike encoders on three popular image and video datasets. The performance of neural encoders working with both deep neural networks (DNNs) and SNNs is analyzed. Our performance data is largely consistent with the benchmark data on image classification from other papers, while limited performance data on the University of Central Florida’s 101 (UCF-101) video dataset were found in comparable studies on spike encoders. Based on our encoder performance data, the Interspike Interval (ISI) encoder performs well across all three datasets, preserving continuous, detailed spike timing and richer temporal information for standard classification tasks. Further, for image classification, multiplexing encoders outperform other spike encoders as they simplify timing patterns by enforcing phase locking and improve stability and robustness to noise. Within the SNN testbenches, the ISI-Phase encoder achieved the highest accuracy on the Modified National Institute of Standards and Technology (MNIST) dataset, surpassing the Time-To-First Spike (TTFS) encoder by 1.9%. On the Canadian Institute For Advanced Research (CIFAR-10) dataset, the ISI encoder achieved the highest accuracy. This ISI encoder had 22.7% higher accuracy than the TTFS encoder on the CIFAR-10 dataset. The ISI encoder performed best on the UCF-101 dataset, achieving 12.7% better performance than the TTFS encoder. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

24 pages, 14412 KB

Open AccessEditor’s ChoiceArticle

Modeling and Trajectory Tracking Control for Double- Steering Wheeled Climbing Robot Based on Adaptive Dynamic Programming

by Zhentao Du, Shiqiang Zhu, Cheng Wang and Wei Song

Electronics 2026, 15(6), 1193; https://doi.org/10.3390/electronics15061193 - 13 Mar 2026

Viewed by 568

Abstract

A model and tracking control method for a double-steering wheeled climbing robot (DSWCR) are presented in this article. The dynamic model of the DSWCR system is established using the Lagrange equation, considering the effects of slipping, variations in gravity and the friction coefficient, [...] Read more.

A model and tracking control method for a double-steering wheeled climbing robot (DSWCR) are presented in this article. The dynamic model of the DSWCR system is established using the Lagrange equation, considering the effects of slipping, variations in gravity and the friction coefficient, and wall/wheel interaction forces. During wall motion, the DSWCR is subject to uncertainties introduced from both the state and model. To address the tracking problem of the DSWCR under state and model uncertainties, an adaptive dynamic programming (ADP) controller based on zero-sum theory is proposed. The stability of the DSWCR tracking system and the convergence of the weights in a neural network are demonstrated. Finally, simulations and a prototype experiment are conducted to verify the optimality and robustness of the proposed control method. Full article

(This article belongs to the Section Systems & Control Engineering)

► Show Figures

Figure 1

19 pages, 764 KB

Open AccessEditor’s ChoiceArticle

FeOCR: Domain-Adaptive Chinese OCR with Visual Character Disambiguation and LLM-Based Correction for Metallurgical Documents

by Qiang Zheng, Yaxuan Sun, Lin Wang, Haoning Zhang, Fanjie Meng and Minghui Li

Electronics 2026, 15(6), 1144; https://doi.org/10.3390/electronics15061144 - 10 Mar 2026

Viewed by 1169

Abstract

High-quality text corpora are essential for knowledge graph construction and domain-specific large model pre-training in technology-intensive industries, with the steel metallurgy sector serving as a representative case. However, many industrial documents remain in scanned or PDF formats, where general-purpose Optical Character Recognition (OCR) [...] Read more.

High-quality text corpora are essential for knowledge graph construction and domain-specific large model pre-training in technology-intensive industries, with the steel metallurgy sector serving as a representative case. However, many industrial documents remain in scanned or PDF formats, where general-purpose Optical Character Recognition (OCR) systems exhibit systematic errors when recognizing Chinese metallurgical documents. In particular, visually similar Chinese characters that differ by only minor strokes are frequently confused, leading to severe degradation of text reliability and cascading errors in downstream knowledge extraction. This paper proposes FeOCR, a general-purpose domain-adaptive framework for machine-printed Chinese characters, which is specifically evaluated within the context of the steel metallurgy industry. The framework integrates visual character disambiguation with context-aware semantic correction. We first construct a metallurgy-specific OCR dataset emphasizing high-frequency confusable Chinese word pairs and enhance data diversity through font perturbation and noise synthesis. Parameter-efficient fine-tuning (LoRA) is then applied to adapt a general OCR model to domain-specific visual patterns. Furthermore, a Large Language Model-based correction module performs semantic refinement of residual errors under domain lexical constraints. Experiments demonstrate significant reductions in character and word error rates, especially for confusable technical terms, providing a reliable foundation for industrial Chinese document digitization. Full article

► Show Figures

Figure 1

39 pages, 1767 KB

Open AccessEditor’s ChoiceSystematic Review

Advanced Hardware Security on Embedded Processors: A 2026 Systematic Review

by Ali Kia, Aaron W. Storey and Masudul Imtiaz

Electronics 2026, 15(5), 1135; https://doi.org/10.3390/electronics15051135 - 9 Mar 2026

Cited by 4 | Viewed by 6519

Abstract

The proliferation of Internet of Things (IoT) devices and embedded processors has recently spurred rapid advances in hardware-level security. This paper systematically reviews developments in securing microcontroller units (MCUs) and constrained embedded platforms from 2020 to 2026, a period marked by the finalization [...] Read more.

The proliferation of Internet of Things (IoT) devices and embedded processors has recently spurred rapid advances in hardware-level security. This paper systematically reviews developments in securing microcontroller units (MCUs) and constrained embedded platforms from 2020 to 2026, a period marked by the finalization of NIST’s post-quantum cryptography standards and accelerated commercial deployment of hardware security primitives. Through analysis of the peer-reviewed literature, industry implementations, and standardization efforts, we survey five critical areas: post-quantum cryptography (PQC) implementations on resource-constrained hardware, physically unclonable functions (PUFs) for device authentication, hardware Roots of Trust and secure boot mechanisms, side-channel attack mitigations, and Trusted Execution Environments (TEEs) for microcontroller-class devices. For each domain, we analyze technical mechanisms, deployment constraints (power, memory, cost), security guarantees, and commercial maturity. Our review distinguishes itself through its integration perspective, examining how these primitives must be composed to secure real-world embedded systems, and its emphasis on post-standardization PQC developments. We highlight critical gaps including PQC memory overhead challenges, ML-resistant PUF designs, and TEE developer friction, while documenting commercial progress such as PSA Level 3 certified components and 500+ million PUF-enabled devices deployed. This synthesis provides practitioners with practical guidance for securing the next generation of IoT and embedded systems. Full article

(This article belongs to the Special Issue Design and Implementation of Embedded Systems for Real-Time Applications)

► Show Figures

Figure 1

20 pages, 1396 KB

Open AccessEditor’s ChoiceArticle

A Cascaded Framework for Vehicle Detection in Low-Resolution Traffic Surveillance Videos

by Tao Yu and Laura Sevilla-Lara

Electronics 2026, 15(5), 1119; https://doi.org/10.3390/electronics15051119 - 8 Mar 2026

Viewed by 686

Abstract

Traffic surveillance cameras, as core sensing devices in smart cities, are crucial for traffic management, violation detection, and autonomous driving. However, due to deployment constraints and hardware limitations, the videos they capture often suffer from low resolution and noise, leading to missed and [...] Read more.

Traffic surveillance cameras, as core sensing devices in smart cities, are crucial for traffic management, violation detection, and autonomous driving. However, due to deployment constraints and hardware limitations, the videos they capture often suffer from low resolution and noise, leading to missed and false detections in traditional object detection algorithms trained on high-resolution data. To address this issue, this study proposes a cascaded collaborative framework that integrates video super-resolution (VSR) and object detection for robust perception in low-quality traffic surveillance scenarios. First, a transformer-based VSR model with masked intra- and inter-frame attention (MIA-VSR) is employed to reconstruct temporally coherent high-resolution video sequences from degraded inputs. A domain-specific super-resolved dataset is subsequently constructed to train a lightweight one-stage detector (You Only Look One-level Feature, YOLOF) for efficient vehicle localisation. Extensive experiments on public datasets (REDS, Vimeo90k, UA-DETRAC) demonstrate that the proposed framework achieved a 56.89 mAP@0.5 on low-resolution UA-DETRAC, outperforming both direct low-resolution inference (39.17 mAP@0.5) and conventional fine-tuning strategies (45.70 mAP@0.5) by 17.72 and 11.19 points, respectively. These findings indicate that super-resolution-driven data reconstruction provides an effective pathway for mitigating feature degradation in low-quality surveillance environments, offering both theoretical insight and practical value for intelligent transportation perception systems. Full article

(This article belongs to the Special Issue Information Systems, Management, and Digital Innovation: Complexity, Integration, and Transformation)

► Show Figures

Figure 1

24 pages, 790 KB

Open AccessEditor’s ChoiceArticle

Maturity-Aware Cyber Insurance Optimization in IoT Networks

by Bishwa Bhusal, Delong Li, Xu Wang and Guangsheng Yu

Electronics 2026, 15(5), 1038; https://doi.org/10.3390/electronics15051038 - 2 Mar 2026

Viewed by 575

Abstract

As the rapid evolution and expansion of Internet of Things (IoT) devices continues to accelerate, modern infrastructures face increasing cyber risks, largely driven by device inter-connectivity, limited security maturity, and interdependent attack propagation across networks. Traditional cyber insurance models often overlook these IoT-specific [...] Read more.

As the rapid evolution and expansion of Internet of Things (IoT) devices continues to accelerate, modern infrastructures face increasing cyber risks, largely driven by device inter-connectivity, limited security maturity, and interdependent attack propagation across networks. Traditional cyber insurance models often overlook these IoT-specific characteristics, relying on uniform or simplified risk assumptions that fail to capture real-world vulnerabilities. To address this gap, this paper presents a maturity-aware cyber insurance optimization framework tailored for interconnected IoT environments. The framework integrates organizational security maturity, interdependent risk propagation modeled through a modified Susceptible–Infected–Susceptible (SIS) process, and a Stackelberg game formulation that captures strategic interactions between the insurer and the defender. Through numerical studies on representative IoT topologies, we demonstrate that maturity-aware, risk-sensitive premium structures quantitatively outperform uniform pricing baselines in cost-efficiency and insurer sustainability. Specifically, our experimental results reveal that operating at an optimal intermediate maturity level (

M = 3

) reduces the defender’s total expected cost by approximately

40 %

(from

255.38

k to

152.36

k) compared to the baseline state (

M = 1

). Furthermore, this structural hardening triggers an

88.3 %

reduction in full-coverage insurance premiums (from

225.38

k to

26.36

k). In contrast, our uniform-pricing baseline exhibits reduced profitability in our experiments due to cross-subsidization effects, reinforcing the value of tiered, risk-proportional pricing for mitigating adverse-selection incentives. In summary, this work establishes a tractable, economically viable framework for cyber insurance in IoT ecosystems and provides a foundation for future extensions to richer network settings. Full article

(This article belongs to the Special Issue Blockchain and Artificial Intelligence for Next-Generation Network and Cybersecurity)

► Show Figures

Figure 1

15 pages, 7961 KB

Open AccessEditor’s ChoiceArticle

A Compact Single-Resonator Dual-Port Circularly Polarized MIMO Dielectric Resonator Antenna for 28 GHz Applications

by Sumer Singh Singhwal and Ladislau Matekovits

Electronics 2026, 15(5), 977; https://doi.org/10.3390/electronics15050977 - 27 Feb 2026

Cited by 1 | Viewed by 664

Abstract

A compact dual-port circularly polarized (CP) multiple-input multiple-output (MIMO) dielectric resonator antenna (DRA) for 28 GHz applications is presented. A single cross-shaped dielectric resonator is excited by two orthogonal microstrip feeds, supporting hybrid orthogonal modes that enable CP radiation at both ports without [...] Read more.

A compact dual-port circularly polarized (CP) multiple-input multiple-output (MIMO) dielectric resonator antenna (DRA) for 28 GHz applications is presented. A single cross-shaped dielectric resonator is excited by two orthogonal microstrip feeds, supporting hybrid orthogonal modes that enable CP radiation at both ports without requiring perturbation cuts, parasitic elements, or decoupling structures. The fabricated prototype exhibits a measured 10 dB impedance bandwidth and 3 dB axial ratio bandwidth that fully cover the Federal Communications Commission (FCC)-allocated 28 GHz band (27.5–28.35 GHz). Port isolation remains better than 15 dB, and the antenna exhibits a peak gain of approximately 7.6 dBi with radiation efficiency exceeding 93%, within a compact 40 × 47 mm² footprint. MIMO performance is verified through envelope correlation coefficient (ECC), diversity gain (DG), and total active reflection coefficient (TARC). The results demonstrate that the proposed single-resonator dual-port CP DRA provides an efficient and integration-friendly solution for compact mmWave MIMO applications in next-generation 5G/6G terminals. Full article

(This article belongs to the Special Issue Next-Generation Antenna Technologies for Wireless Communications and Beyond)

► Show Figures

Figure 1

21 pages, 4260 KB

Open AccessEditor’s ChoiceArticle

CMCLTrack: Reliability-Modulated Cross-Modal Adapter and Cross-Layer Mamba Fusion for RGB-T Tracking

by Pengfei Li, Xiaohe Li and Zide Fan

Electronics 2026, 15(5), 989; https://doi.org/10.3390/electronics15050989 - 27 Feb 2026

Viewed by 749

Abstract

Single-object tracking has progressed rapidly, yet it remains fragile under low illumination, occlusion, and background clutter. RGB-Thermal (RGB-T) tracking improves robustness via modality complementarity, yet many existing trackers do not dynamically switch the dominant modality as sensing quality changes and often rely on [...] Read more.

Single-object tracking has progressed rapidly, yet it remains fragile under low illumination, occlusion, and background clutter. RGB-Thermal (RGB-T) tracking improves robustness via modality complementarity, yet many existing trackers do not dynamically switch the dominant modality as sensing quality changes and often rely on simple late fusion at a single stage, underutilizing multi-level features across the backbone. To address these challenges, we propose CMCLTrack, a unified framework that integrates the Reliability-Modulated Cross-Modal Adapter (RMCA) and the Cross-Layer Mamba Fusion (CLMF). Specifically, RMCA performs reliability-aware bidirectional cross-modal interaction by dynamically weighting modality contributions, while CLMF efficiently aggregates complementary cues from multiple encoder layers to exploit multi-level representations. To stabilize the learning of layer-wise modality reliability, we additionally incorporate a cross-layer reliability smoothness regularization. Extensive experiments on multiple RGB-T tracking benchmarks demonstrate that CMCLTrack achieves competitive performance compared to existing state-of-the-art methods. Full article

(This article belongs to the Special Issue Advances in Multitarget Tracking and Applications)

► Show Figures

Figure 1

35 pages, 1070 KB

Open AccessEditor’s ChoiceArticle

Adaptive Deep Learning Framework for Emotion Recognition in Social Robots: Toward Inclusive Human–Robot Interaction for Users with Special Needs

by Eryka Probierz and Adam Gałuszka

Electronics 2026, 15(5), 924; https://doi.org/10.3390/electronics15050924 - 25 Feb 2026

Viewed by 1019

Abstract

Emotion recognition is a key capability of social robots operating in real-world human-centered environments, especially when interacting with users with special needs. Such users may express emotions in atypical, subtle, or strongly context-dependent ways. These characteristics pose significant challenges for conventional emotion recognition [...] Read more.

Emotion recognition is a key capability of social robots operating in real-world human-centered environments, especially when interacting with users with special needs. Such users may express emotions in atypical, subtle, or strongly context-dependent ways. These characteristics pose significant challenges for conventional emotion recognition systems. This paper proposes an adaptive deep learning framework for emotion recognition in social robots. The framework is designed to support inclusive and accessible human–robot interaction. It combines region-based convolutional neural networks with adaptive learning mechanisms. These mechanisms explicitly model individual variability, contextual information, and interaction dynamics. Multiple deep architectures are evaluated to assess robustness across diverse emotional expressions, including those influenced by cognitive, sensory, or developmental differences. Rather than relying on fixed emotion models, the proposed approach emphasizes adaptability. The system dynamically adjusts its perception strategies to user-specific expressive patterns. Experimental validation is conducted using context-aware emotion datasets. Performance is evaluated in terms of detection accuracy, robustness to variability, and generalization across emotion categories. The results show that adaptive mechanisms improve recognition performance in scenarios characterized by non-standard or low-intensity expressions, compared to static baseline models. This study highlights the importance of flexible, context-sensitive perception for inclusive social robotics. It also discusses design implications for deploying emotion-aware robots in assistive, educational, and therapeutic settings. Overall, the proposed framework represents a step toward socially intelligent robots capable of engaging more effectively with users with special needs. Full article

(This article belongs to the Special Issue Research on Deep Learning and Human-Robot Collaboration)

► Show Figures

Figure 1

20 pages, 3629 KB

Open AccessEditor’s ChoiceArticle

HS-FP and SS-FP: Fine-Pruning-Based Backdoor Elimination for Spiking Neural Networks on Neuromorphic Event Data

by Ki-Ho Kim and Eun-Kyu Lee

Electronics 2026, 15(5), 937; https://doi.org/10.3390/electronics15050937 - 25 Feb 2026

Cited by 1 | Viewed by 730

Abstract

Spiking Neural Networks (SNNs) have attracted increasing attention due to their energy efficiency and suitability for neuromorphic data processing. Despite these advantages, the security of SNNs—particularly their robustness against backdoor attacks—remains underexplored. This study revisits fine-pruning, a widely adopted backdoor defense technique in [...] Read more.

Spiking Neural Networks (SNNs) have attracted increasing attention due to their energy efficiency and suitability for neuromorphic data processing. Despite these advantages, the security of SNNs—particularly their robustness against backdoor attacks—remains underexplored. This study revisits fine-pruning, a widely adopted backdoor defense technique in deep neural networks, and adapts it to the unique spatio-temporal characteristics of SNNs. We propose two SNN-specific fine-pruning methods: Hook–Surrogate Gradient-based fine-pruning (HS-FP) and Spike–STDP-based fine-pruning (SS-FP). HS-FP leverages hook-based activation analysis with surrogate gradient learning, while SS-FP integrates total spike activity with hybrid STDP and surrogate gradient fine-tuning. We evaluate both methods against static, moving, and smart backdoor attacks on two neuromorphic benchmarks, N-MNIST and DVS128-Gesture. Experimental results show that both approaches reduce the attack success rate down to approximately 10% while preserving model accuracy above 99% on N-MNIST and achieving substantial recovery on DVS128-Gesture. Moreover, our analysis reveals that several phenomena observed in fine-pruning-based defenses for deep neural networks—such as mixed-function neurons and backdoor reactivation during fine-tuning—also manifest in SNNs. These findings highlight both the effectiveness and limitations of fine-pruning in the SNN domain and suggest promising directions for extending existing DNN security methodologies to neuromorphic systems. Full article

(This article belongs to the Special Issue Innovative Architectures and Advanced Solutions for Network Security in the Era of Emerging Technologies)

► Show Figures

Figure 1

14 pages, 2793 KB

Open AccessEditor’s ChoiceArticle

A Cross-Domain Authentication Key Agreement Protocol for Edge Computing

by Zhaobo Wang, Wen Feng, Yifeng Yin and Zhiyong Jing

Electronics 2026, 15(5), 946; https://doi.org/10.3390/electronics15050946 - 25 Feb 2026

Viewed by 695

Abstract

With the rapid development of edge computing in the Industrial Internet, data sharing schemes among edge users require reliable cross-domain authentication and key agreement mechanisms to guarantee the security and reliability of inter-device communication. To tackle the deficiencies of existing group key agreement [...] Read more.

With the rapid development of edge computing in the Industrial Internet, data sharing schemes among edge users require reliable cross-domain authentication and key agreement mechanisms to guarantee the security and reliability of inter-device communication. To tackle the deficiencies of existing group key agreement schemes, including dependence on trusted third parties, high computational overhead, and the difficulty of achieving both privacy preservation and attack resistance, this paper presents a cross-domain authenticated key agreement protocol designed for edge computing environments. This protocol supports anonymous identity authentication between cross-domain users, and innovatively constructs a multi-dimensional virtual iterative cyberspace model to generate massive secure keys via the collaborative iteration of multi-user key sequences. The proposed protocol is decentralized, lightweight, and resistant to replay attacks and man-in-the-middle attacks, while satisfying forward and backward secrecy. Security analysis and performance comparison experiments illustrate that the protocol significantly reduces computational and communication overhead, matches the resource-constrained characteristics of edge devices, and can be widely deployed in large-scale data encryption and sharing scenarios under edge computing environments. Full article

(This article belongs to the Section Networks)

► Show Figures

Figure 1

22 pages, 1795 KB

Open AccessEditor’s ChoiceArticle

PPO-Based Hybrid Optimization for RIS-Assisted Semantic Vehicular Edge Computing

by Wei Feng, Jingbo Zhang, Qiong Wu, Pingyi Fan and Qiang Fan

Electronics 2026, 15(5), 936; https://doi.org/10.3390/electronics15050936 - 25 Feb 2026

Viewed by 623

Abstract

To support latency-sensitive Internet of Vehicles (IoV) applications amidst dynamic environments and intermittent links, this paper proposes a Reconfigurable Intelligent Surface (RIS)-aided semantic-aware Vehicle Edge Computing (VEC) framework. This approach integrates RIS to optimize wireless connectivity and semantic communication to minimize latency by [...] Read more.

To support latency-sensitive Internet of Vehicles (IoV) applications amidst dynamic environments and intermittent links, this paper proposes a Reconfigurable Intelligent Surface (RIS)-aided semantic-aware Vehicle Edge Computing (VEC) framework. This approach integrates RIS to optimize wireless connectivity and semantic communication to minimize latency by transmitting semantic features. We formulate a comprehensive joint optimization problem by optimizing offloading ratios, the number of semantic symbols, and RIS phase shifts. Considering the problem’s high dimensionality and non-convexity, we propose a two-tier hybrid scheme that employs Proximal Policy Optimization (PPO) for discrete decision-making and Linear Programming (LP) for offloading optimization. The simulation results have validated the proposed framework’s superiority over existing methods. Specifically, the proposed PPO-based hybrid optimization scheme reduces the average end-to-end latency by approximately 40% to 50% compared to Genetic Algorithm (GA) and Quantum-behaved Particle Swarm Optimization (QPSO). Moreover, the system demonstrates strong scalability by maintaining low latency even in congested scenarios with up to 30 vehicles. Full article

(This article belongs to the Section Microwave and Wireless Communications)

► Show Figures

Graphical abstract

15 pages, 6200 KB

Open AccessEditor’s ChoiceArticle

A Beam-Splitter-Free Terahertz Receiver with Independent Antenna-Fed Local Oscillator for Enhanced Efficiency

by Pengfei Zhao, Dabao Wang, Xinyu Yao, Ning Liu, Xiaochun Jiao and Jing Cao

Electronics 2026, 15(5), 919; https://doi.org/10.3390/electronics15050919 - 24 Feb 2026

Viewed by 470

Abstract

This paper presents the design, fabrication, and experimental characterization of a novel terahertz receiver comprising two high-performance receiving antennas and a combiner. The low efficiency of local oscillator (LO) power utilization, caused by conventional beam splitters, presents a major bottleneck for large-array terahertz [...] Read more.

This paper presents the design, fabrication, and experimental characterization of a novel terahertz receiver comprising two high-performance receiving antennas and a combiner. The low efficiency of local oscillator (LO) power utilization, caused by conventional beam splitters, presents a major bottleneck for large-array terahertz receivers. By eliminating the conventional beam splitter, the proposed system allows the terahertz signal and LO power to be directly and independently received by two dedicated antennas, thereby significantly enhancing LO power efficiency. The receiver is successfully fabricated using micromachining technology into a compact 2.5-dimensional multilayered structure measuring 9 mm × 16 mm × 7.2 mm. Key performance metrics, including the waveguide port S-parameters, radiation patterns, and gains of the two horn antennas, were measured. The experimental results show close agreement with simulations, validating the system’s accuracy and reliability. Furthermore, the system’s equivalent noise temperature was measured to be 395 K, indicating excellent thermal stability and sensitivity. This study concludes that the proposed terahertz receiver design is both feasible and efficient for high-resolution applications, showing great potential for use in satellite-based space observation systems or base stations requiring advanced terahertz signal processing. Full article

(This article belongs to the Section Microwave and Wireless Communications)

► Show Figures

Figure 1

26 pages, 9016 KB

Open AccessFeature PaperEditor’s ChoiceArticle

Integration of Hybrid Prefilter and Corner Trajectory Planning for Simultaneously Suppressing Residual Vibration and Reducing Cornering Error of SCARA Robots

by Syh-Shiuh Yeh and Ming-Han You

Electronics 2026, 15(4), 900; https://doi.org/10.3390/electronics15040900 - 23 Feb 2026

Viewed by 590

Abstract

During high-speed cornering, the motion accuracy and efficiency of SCARA robots are often compromised by residual vibrations and cornering errors. Conventional control methods often fail to address these two coupled problems simultaneously. Therefore, this study developed an integrated design strategy to simultaneously suppress [...] Read more.

During high-speed cornering, the motion accuracy and efficiency of SCARA robots are often compromised by residual vibrations and cornering errors. Conventional control methods often fail to address these two coupled problems simultaneously. Therefore, this study developed an integrated design strategy to simultaneously suppress residual vibrations and restrict cornering errors for improving the cornering performance of the SCARA robot. The core of this design strategy is to develop a hybrid prefilter via the convolution of an input shaper and a finite impulse response filter, thereby creating a prefilter with robust, high-performance residual vibration suppression. Subsequently, to accommodate the asymmetric acceleration and deceleration generated by the hybrid prefilter, this study developed a systematic corner trajectory planning method that can calculate the cornering trajectory parameters based on a preset value of the cornering error to restrict the cornering error and ensure the cornering accuracy of the SCARA robot. Experimental results indicated that under the condition of a restricted cornering error, the developed hybrid prefilter can reduce residual vibration by >85%. Thus, the hybrid prefilter designed with the corner trajectory planning method can mitigate the coupled problem of residual vibration and cornering error, suppressing the residual vibration without compromising cornering accuracy. Full article

(This article belongs to the Special Issue Intelligent Perception and Control for Robotics)

► Show Figures

Figure 1

21 pages, 4060 KB

Open AccessEditor’s ChoiceArticle

Machine Learning and Regression-Based Multimodal Intelligent Injury Severity Modeling of Median Crossover Crashes

by Deo Chimba, Sandeep Bist, Jeannine Mbabazi, Philbert Mwandepa and Wittness Mariki

Electronics 2026, 15(4), 901; https://doi.org/10.3390/electronics15040901 - 23 Feb 2026

Viewed by 715

Abstract

Median crossover crashes are among the most severe roadway safety events due to their high-energy nature and strong association with fatal and incapacitating injuries, posing a substantial public health burden. This study develops a multimodal intelligent analytics framework to evaluate the cable median [...] Read more.

Median crossover crashes are among the most severe roadway safety events due to their high-energy nature and strong association with fatal and incapacitating injuries, posing a substantial public health burden. This study develops a multimodal intelligent analytics framework to evaluate the cable median barrier performance in Tennessee by integrating structured crash data, roadway and traffic characteristics, post-impact vehicle responses, and unstructured police narratives. Across 6094 crashes on 576 cable barrier segments, 1196 involved barrier impacts and 914 included complete post-impact response information. Deep learning-based text mining using a BERT transformer model was applied to narrative descriptions from fatal, serious injury, and minor injury crashes to extract contextual indicators of loss of control, impact dynamics, and injury mechanisms. Safety effectiveness evaluation using Empirical Bayes methods showed substantial reductions after installation, including a 96% decrease in fatal crashes and an 88% reduction in serious-injury crashes. Vehicle–barrier interactions—classified as containment, redirection, rollover, or penetration—were modeled using a multinomial logit framework with marginal effects to assess the influence of geometric, operational, and vehicle-related factors. Reduced barrier offset, narrow shoulders, high traffic volumes, outer-lane departures, and heavy-vehicle involvement significantly increased the likelihood of rollover and penetration events, which are strongly linked to higher injury severity. Through fusing multimodal data and combining explainable statistical models with deep learning text analysis, this study provided a scalable, trustworthy approach to characterizing injury risk, aligning transportation safety analytics with emerging intelligent healthcare and big-data methodologies aimed at preventing severe and fatal trauma. Full article

(This article belongs to the Special Issue Multimodal Intelligent Healthcare and Big Data Analysis)

► Show Figures

Graphical abstract

22 pages, 1271 KB

Open AccessEditor’s ChoiceArticle

Leveraging MCP and Corrective RAG for Scalable and Interoperable Multi-Agent Healthcare Systems

by Dimitrios Kalathas, Andreas Menychtas, Panayiotis Tsanakas and Ilias Maglogiannis

Electronics 2026, 15(4), 888; https://doi.org/10.3390/electronics15040888 - 21 Feb 2026

Viewed by 1641

Abstract

The rapid evolution of Generative AI (GenAI) has created the conditions for developing innovative solutions that disrupt all fields of human-related activities. Within the healthcare sector, numerous AI-driven applications have emerged, offering comprehensive health-related insights and addressing user questions in real time. Nevertheless, [...] Read more.

The rapid evolution of Generative AI (GenAI) has created the conditions for developing innovative solutions that disrupt all fields of human-related activities. Within the healthcare sector, numerous AI-driven applications have emerged, offering comprehensive health-related insights and addressing user questions in real time. Nevertheless, most of them use general-purpose Large Language Models (LLMs); consequently, the responses may not be as accurate as required in clinical settings. Therefore, the research community is adopting efficient architectures, such as Multi-Agent Systems (MAS) to optimize task allocation, reasoning processes, and system scalability. Most recently, the Model Context Protocol (MCP) has been introduced; however, very few applications apply this protocol within a healthcare MAS. Furthermore, Retrieval-Augmented Generation (RAG) has proven essential for grounding AI responses in verified clinical literature. This paper proposes a novel architecture that integrates these technologies to create an advanced Agentic Corrective RAG (CRAG) system. Unlike standard approaches, this method incorporates an active evaluation layer that autonomously detects retrieval failures and triggers corrective fallback mechanisms to ensure safety and accuracy. A comparative analysis was conducted for this architecture against Typical RAG and Cache-Augmented Generation (CAG), demonstrating that the proposed solution improves workflow efficiency and enables more accurate, context-aware interventions in healthcare. Full article

(This article belongs to the Special Issue Artificial Intelligence-Driven Emerging Applications)

► Show Figures

Figure 1

16 pages, 1038 KB

Open AccessFeature PaperEditor’s ChoiceArticle

The Agency-First Framework: Operationalizing Human-Centric Interaction and Evaluation Heuristics for Generative AI

by Christos Troussas, Christos Papakostas, Akrivi Krouska and Cleo Sgouropoulou

Electronics 2026, 15(4), 877; https://doi.org/10.3390/electronics15040877 - 20 Feb 2026

Viewed by 1966

Abstract

Current generative AI systems primarily utilize a prompt–response interaction model that restricts user intervention during the creative process. This lack of granular control creates a significant disconnect between user intent and machine output, which we define as the “Agency Gap”. This paper introduces [...] Read more.

Current generative AI systems primarily utilize a prompt–response interaction model that restricts user intervention during the creative process. This lack of granular control creates a significant disconnect between user intent and machine output, which we define as the “Agency Gap”. This paper introduces the Agency-First Framework (AFF), which combines cognitive engineering and co-active design approaches to formally define human-AI collaboration. This is operationalized through the development of ten Generative AI Agency (GAIA) Heuristics, a systematic method for evaluating agency-centric interactions within stochastic generative settings. By translating the theoretical layers of the AFF into measurable criteria, the GAIA heuristics provide the necessary instrument for the empirical auditing of existing systems and the guidance of agency-centric redesigns. Unlike existing assistive AI guidelines that focus on output-level usability, the AFF establishes agency as a first-class design construct, enabling mid-process intervention and the steering of the model’s latent reasoning trajectory. Validation of the AFF was conducted through a two-tiered empirical evaluation: (1) an expert heuristic audit of state-of-the-art platforms, such as ChatGPT-o1 and Midjourney v6, which achieved high inter-rater reliability, and (2) a controlled redesign study. The latter demonstrated that agency-centric interfaces significantly enhance the Sense of Agency and Intent Alignment Accuracy compared to baseline prompt-response models, even when introducing a deliberate increase in task completion time—a phenomenon we describe as “productive friction” or an intentional interaction slowdown designed to prioritize cognitive engagement and user control over raw speed. Overall, the findings suggest that the restoration of meaningful user agency requires a shift from “seamless” system efficiency towards “productive friction”, where controllability and transparency within the generative process are prioritized. The major contribution of this work is the provision of a scalable, empirically validated framework and set of heuristics that equip designers to move beyond prompt-centric interaction, establishing a methodological foundation for agency-preserving generative AI systems. Full article

(This article belongs to the Special Issue The Role of Artificial Intelligence in Enhancing Human-Computer Interaction)

► Show Figures

Figure 1

16 pages, 2588 KB

Open AccessFeature PaperEditor’s ChoiceArticle

Smart Home IoT Forensics in Matter Ecosystems: A Data Extraction Method Using Multi-Admin

by Sungbum Kim, Sungmoon Kwon and Taeshik Shon

Electronics 2026, 15(4), 884; https://doi.org/10.3390/electronics15040884 - 20 Feb 2026

Cited by 1 | Viewed by 992

Abstract

As the smart home ecosystem expands with the adoption of Matter, a wide variety of Internet of Things (IoT) devices are entering the market, and these devices are becoming more complex, as they support diverse functionalities. Consequently, smart home forensics often requires data [...] Read more.

As the smart home ecosystem expands with the adoption of Matter, a wide variety of Internet of Things (IoT) devices are entering the market, and these devices are becoming more complex, as they support diverse functionalities. Consequently, smart home forensics often requires data extraction procedures that are specific to each device and platform, which increases the technical burden and time costs for investigators. To address these challenges, this study proposes a method that leverages Matter Multi-Admin support for multiple fabrics to enable efficient data acquisition from Matter-enabled IoT devices, regardless of the underlying smart home platform. This method configures a forensic Matter controller using chip-tool and commissions IoT devices that have already been commissioned to a smart home platform into a secondary fabric via Multi-Admin. The forensic controller then performs data extraction using standardized Matter interfaces. The proposed approach was validated on our smart home testbed by targeting a Matter smart bulb commissioned to the SmartThings platform and successfully extracting data generated by the platform, thereby demonstrating the utility of the method. The results indicate that the method enables nondestructive and efficient evidence acquisition from smart home IoT devices and can support future research and real-world investigations. Full article

(This article belongs to the Special Issue New Challenges in IoT Security)

► Show Figures

Figure 1

30 pages, 8046 KB

Open AccessFeature PaperEditor’s ChoiceArticle

A Progressive Evaluation of MIMO Techniques in LoRa-Type Wireless Sensor Networks Under Imperfect Channel State Information

by Nikolaos Mouziouras, Andreas Tsormpatzoglou and Constantinos T. Angelis

Electronics 2026, 15(4), 867; https://doi.org/10.3390/electronics15040867 - 19 Feb 2026

Cited by 1 | Viewed by 563

Abstract

Low-Power Wide-Area Network (LPWAN) technologies play a central role in large-scale wireless sensor network (WSN) deployments, where energy efficiency, coverage and reliability dominate over throughput. Among them, Long Range (LoRa) technology has emerged as a widely adopted physical-layer solution due to its ability [...] Read more.

Low-Power Wide-Area Network (LPWAN) technologies play a central role in large-scale wireless sensor network (WSN) deployments, where energy efficiency, coverage and reliability dominate over throughput. Among them, Long Range (LoRa) technology has emerged as a widely adopted physical-layer solution due to its ability to operate at extremely low signal-to-noise ratios (SNRs). While multi-antenna techniques can potentially enhance link performance, their applicability in LoRa-type systems is constrained by low-SNR operation, strict energy budgets and the quality of channel state information (CSI). This paper presents a systematic and progressively structured evaluation of multiple-input multiple-output (MIMO) techniques in LoRa-type systems under representative operating conditions. A multi-stage simulation framework, implemented using the Vienna SLS v2.0 (Q3) simulator and adapted to LoRa-like waveforms, is employed to isolate the impact of large-scale propagation, small-scale fading, antenna configuration and CSI quality. The analysis starts from a system-level coverage baseline and advances to link-level evaluations of diversity-oriented MIMO schemes and spatial multiplexing configurations under both ideal and imperfect CSI. The results demonstrate that spatial diversity techniques are well aligned with the operational characteristics of LoRa links, offering robust performance in low-SNR regimes and under limited CSI accuracy. In contrast, spatial multiplexing exhibits higher sensitivity to channel estimation errors, with its practical benefits becoming apparent primarily when evaluated using throughput-oriented metrics such as packet error rate and normalized goodput. Overall, the study highlights the fundamental trade-off between reliability and capacity in LoRa MIMO systems and provides design-oriented insights for wireless sensor network deployments. Full article

(This article belongs to the Special Issue Wireless Sensor Network: Latest Advances and Prospects)

► Show Figures

Figure 1

42 pages, 1277 KB

Open AccessEditor’s ChoiceArticle

A Hybrid Time Series Forecasting Model Combining ARIMA and Decision Trees to Detect Attacks in MITRE ATT&CK Labeled Zeek Log Data

by Raymond Freeman, Sikha S. Bagui, Subhash C. Bagui, Dustin Mink, Sarah Cameron and Germano Correa Silva De Carvalho

Electronics 2026, 15(4), 871; https://doi.org/10.3390/electronics15040871 - 19 Feb 2026

Cited by 1 | Viewed by 1061

Abstract

Intrusion detection systems face challenges in processing high-volume network traffic while maintaining accuracy across diverse low volume attack types. This study presents a hybrid approach combining ARIMA time series forecasting with Decision Tree classification to detect attacks in Zeek network flow data labeled [...] Read more.

Intrusion detection systems face challenges in processing high-volume network traffic while maintaining accuracy across diverse low volume attack types. This study presents a hybrid approach combining ARIMA time series forecasting with Decision Tree classification to detect attacks in Zeek network flow data labeled with MITRE ATT&CK tactics, leveraging PySpark for scalability. ARIMA identifies temporal anomalies which Decision Trees then classify by attack type. The ARIMA model was evaluated across 13 MITRE ATT&CK tactics, though only 7 maintained sufficient class balance for valid assessment. Results are reported at three evaluation levels: Baseline (Decision Tree only), ARIMA-DT (Decision Tree tested on ARIMA-filtered anomalies), and End-to-End (pipeline performance measured against the original test population). The hybrid model demonstrated two distinct benefits: performance improvement for detectable attacks and detection enablement for previously undetectable attacks. For high-volume attacks with existing baseline detection, ARIMA preprocessing substantially improved performance, for example, Reconnaissance achieved an ARIMA-DT F1 score of 99.71% (from a baseline of 80.88%) with End-to-End metrics confirming this improvement at 97.59% F1-score. Credential Access reached a perfect 100% precision and recall on the ARIMA-filtered subset (from a baseline recall of 7.48%); however, End-to-End evaluation revealed that ARIMA filtering removed the vast majority of Credential Access attacks, resulting in a 1.28% End-to-End F1-score—worse than the baseline F1-score of 7.41%—demonstrating that the hybrid pipeline is counterproductive for attack types whose flow characteristics closely resemble legitimate traffic. More significantly, ARIMA preprocessing enabled detection where traditional Decision Trees completely failed (0% recall) for four stealthy attack types: Defense Evasion (ARIMA-DT recall of 93.22%, End-to-End 67.83%), Discovery (ARIMA-DT recall of 100%, End-to-End 63.43%), Persistence (ARIMA-DT recall of 86.92%, End-to-End 73.38%), and Privilege Escalation (ARIMA-DT recall of 89.93%, End-to-End 64.68%). These results demonstrate that ARIMA-based statistical anomaly detection is particularly effective for attacks involving subtle, low-volume activities that blend with legitimate operations, while also improving classification accuracy for high-volume reconnaissance activities. Full article

(This article belongs to the Special Issue Recent Advances in Intrusion Detection Systems Using Machine Learning)

► Show Figures

Figure 1

25 pages, 1932 KB

Open AccessEditor’s ChoiceArticle

Blockchain-Enabled Governance for Health IoT Data Access via Interpretable Multi-Objective Optimization and Bargaining Under Privacy–Latency–Robustness Trade-Offs

by Farshid Keivanian, Yining Hu and Saman Shojae Chaeikar

Electronics 2026, 15(4), 864; https://doi.org/10.3390/electronics15040864 - 18 Feb 2026

Viewed by 896

Abstract

Health Internet of Things (Health IoT) systems continuously stream sensitive physiological data, making data access governance safety-critical under conflicting objectives such as privacy risk, latency, energy/resource cost, and robustness, especially when conditions change during emergencies. This paper proposes FiB-MOBA-EAFG, a hybrid blockchain–AI framework [...] Read more.

Health Internet of Things (Health IoT) systems continuously stream sensitive physiological data, making data access governance safety-critical under conflicting objectives such as privacy risk, latency, energy/resource cost, and robustness, especially when conditions change during emergencies. This paper proposes FiB-MOBA-EAFG, a hybrid blockchain–AI framework that separates on-chain accountability from off-chain decision intelligence. Off-chain, fuzzy context inference parameterizes scenario priorities, Pareto-based multi-objective search generates candidate governance policies, an emergency-aware feasibility guard filters unsafe trade-offs, and a bargaining-based selector chooses a single deployable policy. On chain, the blockchain layer records consent commitments, access events, and hashes of the selected policy and decision trace, serving as an immutable audit and accountability substrate rather than an online decision or optimization engine, while raw health data remain off-chain. Using simulation studies of home remote monitoring, clinic telehealth, and emergency triage under stochastic network variation and adversarial device behavior, FiB-MOBA-EAFG improves robustness and yields more repeatable policy selection than rule-based control and scalarized baselines within the evaluated simulation scenarios, while maintaining latency within ranges compatible with modeled edge deployment constraints through explicit emergency-aware feasibility constraints. A budget-matched random-search ablation further indicates that structured Pareto exploration is needed to reliably obtain robust, low-risk governance policies. Full article

(This article belongs to the Special Issue Blockchain-Enabled Management Systems in Health IoT)

► Show Figures

Figure 1

22 pages, 2506 KB

Open AccessEditor’s ChoiceArticle

CycleGAN-Based Data Augmentation for Scanning Electron Microscope Images to Enhance Integrated Circuit Manufacturing Defect Classification

by Andrew Yen, Nemo Chang, Jean Chien, Lily Chuang and Eric Lee

Electronics 2026, 15(4), 803; https://doi.org/10.3390/electronics15040803 - 13 Feb 2026

Viewed by 824

Abstract

Semiconductor defect inspection is frequently hindered by data scarcity and the resulting class imbalance in supervised learning. This study proposes a CycleGAN-based data augmentation pipeline designed to synthesize realistic defective CD-SEM images from abundant normal patterns, incorporating a quantitative quality control mechanism. Using [...] Read more.

Semiconductor defect inspection is frequently hindered by data scarcity and the resulting class imbalance in supervised learning. This study proposes a CycleGAN-based data augmentation pipeline designed to synthesize realistic defective CD-SEM images from abundant normal patterns, incorporating a quantitative quality control mechanism. Using an ADI CD-SEM dataset, we conducted a sensitivity analysis by cropping original 1024 × 1024 micrographs into 512 × 512 and 256 × 256 inputs. Our results indicate that increasing the effective defect-area ratio is critical for improving generative stability and defect visibility. To ensure data integrity, we applied a screening protocol based on the Structural Similarity Index (SSIM) and a median absolute deviation noise metric to exclude low-fidelity outputs. When integrated into the training of XceptionNet classifiers, this filtered augmentation strategy yielded substantial performance gains on a held-out test set, specifically improving the Recall and F1 score while maintaining a near-ceiling AUC. These results demonstrate that controlled CycleGAN augmentation, coupled with objective quality filtering, effectively mitigates class imbalance constraints and significantly enhances the robustness of automated defect detection. Full article

(This article belongs to the Special Issue Deep Learning in Video and Image Processing: Challenges, Solutions, and Future Directions, 2nd Edition)

► Show Figures

Figure 1

28 pages, 2899 KB

Open AccessFeature PaperEditor’s ChoiceArticle

Design of Secure Communication Networks for UAV Platform Empowered by Lightweight Authentication Protocols

by Muhammet A. Sen, Saba Al-Rubaye and Antonios Tsourdos

Electronics 2026, 15(4), 785; https://doi.org/10.3390/electronics15040785 - 12 Feb 2026

Cited by 1 | Viewed by 936

Abstract

Flying Ad Hoc Networks (FANETs) formed by cooperative Unmanned Aerial Vehicles (UAVs) require formally proven secure and resource-efficient authentication because open wireless channels allow active adversaries to inject commands, replay traffic, and impersonate nodes. Conventional certificate-based mechanisms impose key management overhead and remain [...] Read more.

Flying Ad Hoc Networks (FANETs) formed by cooperative Unmanned Aerial Vehicles (UAVs) require formally proven secure and resource-efficient authentication because open wireless channels allow active adversaries to inject commands, replay traffic, and impersonate nodes. Conventional certificate-based mechanisms impose key management overhead and remain vulnerable under device capture, while existing lightweight and Physical Unclonable Function (PUF)-assisted proposals commonly assume stable connectivity, lack formal adversarial verification, or are evaluated only through simulation. This paper presents a lightweight PUF-assisted authentication protocol designed for dynamic multi-hop FANET operation. The scheme provides mutual UAV–Ground Station (GS) authentication and session key establishment and further enables secure UAV–UAV communication using an off-path ticket mechanism that eliminates continuous infrastructure dependence. The protocol is constructed through verification-driven refinement and formally analysed under the Dolev–Yao model, establishing authentication and session key secrecy and resistance to replay and impersonation attacks. Implementation-oriented latency measurements on Raspberry-Pi-class embedded platforms demonstrate that cryptographic processing time can be further reduced with hardware improvements, while the overall end-to-end delay is still largely determined by channel conditions and connection behaviour. Comparative evaluation shows reduced communication cost and broader security coverage relative to existing UAV authentication schemes, indicating practical deployability in large-scale FANET environments. Full article

(This article belongs to the Special Issue Wireless Sensor Network: Latest Advances and Prospects)

► Show Figures

Graphical abstract

16 pages, 46885 KB

Open AccessEditor’s ChoiceArticle

Monolithic Integration of a Dual-Mode On-Chip Antenna with a Ferroelectric Hafnium Zirconium Oxide Varactor for Reprogrammable Radio-Frequency Front Ends

by Samuel Quaresima, Nicolas Casilli, Sherif Badran, Onurcan Kaya, Vitaly Petrov, Luca Colombo, Benyamin Davaji, Josep Miquel Jornet and Cristian Cassella

Electronics 2026, 15(4), 792; https://doi.org/10.3390/electronics15040792 - 12 Feb 2026

Viewed by 960

Abstract

In this work, we report a dual-mode ferroelectrically programmable on-chip antenna. The antenna is built on a silicon wafer using complementary metal-oxide semiconductor (CMOS) processes and exhibits two programmable resonant modes: one in the super high frequency (SHF) range and one in the [...] Read more.

In this work, we report a dual-mode ferroelectrically programmable on-chip antenna. The antenna is built on a silicon wafer using complementary metal-oxide semiconductor (CMOS) processes and exhibits two programmable resonant modes: one in the super high frequency (SHF) range and one in the extremely high frequency (EHF) range. The SHF mode resonates at 8.5 GHz and exhibits ultrawideband (UWB) behavior, while the EHF mode resonates at 36.6 GHz. Both resonance frequencies can be tuned in a non-volatile fashion by controlling the ferroelectric polarization state of a Hafnium Zirconium Oxide (HZO) varactor monolithically integrated into the feed line. This programmability arises from the ferroelectric switching of the embedded HZO film, which results in a non-volatile variation of its permittivity upon application of a voltage pulse. Ferroelectric switching occurs at approximately ±3 V and induces maximum resonance frequency shifts of 381 MHz for the SHF mode and 3 GHz for the EHF mode, corresponding to fractional frequency changes of 4.5% and 8.3%, respectively. Unlike previously reported ferroelectrically tunable antennas, our reported antenna combines full integration, CMOS compatibility, higher operating frequency, compact footprint, and non-volatile programmability. Full article

(This article belongs to the Special Issue Feature Papers in 'Microwave and Wireless Communications' Section, 2nd Edition)

► Show Figures

Graphical abstract

16 pages, 1372 KB

Open AccessEditor’s ChoiceArticle

Spatio-Temporal Deep Learning-Assisted Multi-Period AC Optimal Power Flow

by Jihun Kim, Sojin Park, Dongwoo Kang and Hunyoung Shin

Electronics 2026, 15(4), 761; https://doi.org/10.3390/electronics15040761 - 11 Feb 2026

Cited by 1 | Viewed by 696

Abstract

The increasing penetration of renewable energy resources has amplified variability and uncertainty in power systems, reducing the effectiveness of conventional single-period Optimal Power Flow (OPF) strategies. Multi-period AC-OPF offers a more comprehensive framework by incorporating inter-temporal constraints and resource flexibility, but its high [...] Read more.

The increasing penetration of renewable energy resources has amplified variability and uncertainty in power systems, reducing the effectiveness of conventional single-period Optimal Power Flow (OPF) strategies. Multi-period AC-OPF offers a more comprehensive framework by incorporating inter-temporal constraints and resource flexibility, but its high computational complexity and strong temporal coupling make large-scale applications challenging, often causing scalability issues and convergence difficulties in conventional solvers. We address these issues with a spatio-temporal deep learning model that combines a Graph Attention Network (GAT) for topology-aware feature learning with a Temporal Convolutional Network (TCN) for multi-period temporal modeling. The proposed model is trained on large-scale 500-bus and 1354-bus systems under both 8-period and 24-period settings, and it achieves robust scalability with consistently high prediction accuracy. Using the model’s predictions, we construct an initial solution and provide it to a conventional OPF solver, which improves convergence performance and demonstrates the model’s effectiveness as an auxiliary tool for complex MP-ACOPF problems. Full article

(This article belongs to the Special Issue Edge-Intelligent Sustainable Cyber-Physical Systems)

► Show Figures

Figure 1

19 pages, 3302 KB

Open AccessEditor’s ChoiceArticle

Empirical Analysis of Heterogeneous Multi-Orbit Satellite Networks for Communication Resilience in Island Regions

by Yi-Cheng Lin, Tuck Wai Choong, Zheng Cheng Pang, Ping-Hsiang Chuang, Yao-Ching Huang, Ming-Te Chen and Jenq-Shiou Leu

Electronics 2026, 15(4), 773; https://doi.org/10.3390/electronics15040773 - 11 Feb 2026

Viewed by 813

Abstract

Integrating Geostationary (GEO), Medium Earth Orbit (MEO), and Low Earth Orbit (LEO) satellite systems offers a promising solution for enhancing communication resilience in disaster-prone island regions. However, effective integration via Software-Defined Wide Area Networks (SD-WANs) faces challenges due to the heterogeneous stochastic characteristics [...] Read more.

Integrating Geostationary (GEO), Medium Earth Orbit (MEO), and Low Earth Orbit (LEO) satellite systems offers a promising solution for enhancing communication resilience in disaster-prone island regions. However, effective integration via Software-Defined Wide Area Networks (SD-WANs) faces challenges due to the heterogeneous stochastic characteristics of these links. This study presents a comprehensive performance benchmark of GEO, MEO, and LEO satellite links based on long-duration empirical campaigns conducted in Taiwan. Our findings quantify critical integration hurdles, specifically the “long-tail” latency distribution in LEO links induced by frequent handovers and significant TCP throughput degradation modeled by the Mathis equation. Furthermore, empirical tests demonstrate that simplistic link aggregation across these heterogeneous orbits results in severe packet reordering and goodput collapse. Based on these results, we propose a conceptual resilience-oriented SD-WAN architecture incorporating intelligent failover thresholds and application-aware routing policies. This work provides foundational data and a design framework to guide the future development of robust multi-layered satellite communication systems for disaster management. Full article

(This article belongs to the Section Networks)

► Show Figures

Figure 1

28 pages, 635 KB

Open AccessEditor’s ChoiceArticle

Harmonizing Supervised Fine-Tuning and Reinforcement Learning with Reward-Based Sampling for Continual Machine Unlearning

by Jiaqi Lang, Jiahao Zhao, Linjing Li and Daniel Dajun Zeng

Electronics 2026, 15(4), 771; https://doi.org/10.3390/electronics15040771 - 11 Feb 2026

Viewed by 1165

Abstract

Large language models (LLMs) are pretrained on massive internet data and inevitably memorize sensitive or copyrighted content. This continually raises privacy, legal, and security concerns. Machine unlearning has been proposed as an approach to remove the influence of undesired data while maintaining model [...] Read more.

Large language models (LLMs) are pretrained on massive internet data and inevitably memorize sensitive or copyrighted content. This continually raises privacy, legal, and security concerns. Machine unlearning has been proposed as an approach to remove the influence of undesired data while maintaining model utility. However, in real-world scenarios, unlearning requests continuously emerge, and existing approaches often struggle to handle these sequential requests, leading to utility degradation. To address this challenge, we propose the harmonization of Supervised fine-tuning and Reinforcement learning with Reward-based Sampling (SRRS) framework, which dynamically harmonizes supervised fine-tuning (SFT) and reinforcement learning (RL) via reward signals: SFT ensures forgetting efficacy, while RL preserves utility under continual adaptation. By harmonizing these paradigms, SRRS achieves reliable forgetting and sustained utility across sequential unlearning tasks, demonstrating competitive performance compared to baseline methods on TOFU and R-TOFU datasets. Full article

(This article belongs to the Special Issue Artificial Intelligence Safety and Security)

► Show Figures

Figure 1

20 pages, 554 KB

Open AccessEditor’s ChoiceArticle

Balancing Long–Short-Term User Preferences via Multilevel Sequential Patterns for Review-Aware Recommendation

by Li Jin, Xinzhe Li, Suji Kim and Jaekyeong Kim

Electronics 2026, 15(4), 753; https://doi.org/10.3390/electronics15040753 - 10 Feb 2026

Viewed by 708

Abstract

Personalized recommender systems play an essential role in enhancing user experience by accurately predicting user preferences. Previous approaches mainly focus on modeling long-term preferences or capturing short-term dynamics through sequential patterns, while few achieve an effective balance between the two. This study proposes [...] Read more.

Personalized recommender systems play an essential role in enhancing user experience by accurately predicting user preferences. Previous approaches mainly focus on modeling long-term preferences or capturing short-term dynamics through sequential patterns, while few achieve an effective balance between the two. This study proposes Rec-SSP, a novel review-aware recommendation model that integrates long-term and short-term preferences through a gated fusion mechanism. Long-term preferences are extracted from aggregated user reviews, whereas short-term preferences are modeled by identifying sequential patterns from recent interactions at both the review and category levels. This multilevel design captures fine-grained opinions across items, ensuring a more accurate understanding of the evolving user intent. This study conducted various experiments on real-world datasets, showing that Rec-SSP outperforms baseline models. These findings demonstrate that balancing long-term and short-term preferences with multilevel sequence modeling can significantly improve recommendation accuracy across diverse domains. Full article

(This article belongs to the Special Issue Machine/Deep Learning Applications and Intelligent Systems)

► Show Figures

Figure 1

36 pages, 24812 KB

Open AccessEditor’s ChoiceReview

Artificial Intelligence-Enhanced Droop Control for Renewable Energy-Based Microgrids: A Comprehensive Review

by Michael Addai and Petr Musilek

Electronics 2026, 15(3), 707; https://doi.org/10.3390/electronics15030707 - 6 Feb 2026

Cited by 5 | Viewed by 1863

Abstract

The integration of renewable energy sources into modern power systems requires advanced control strategies to maintain stability, reliability, and efficiency. This paper presents a comprehensive review of the application of artificial intelligence techniques, including machine learning, deep learning, and reinforcement learning, in improving [...] Read more.

The integration of renewable energy sources into modern power systems requires advanced control strategies to maintain stability, reliability, and efficiency. This paper presents a comprehensive review of the application of artificial intelligence techniques, including machine learning, deep learning, and reinforcement learning, in improving droop control for renewable energy integration. These artificial intelligence-based methods address key challenges such as frequency and voltage regulation, power sharing, and grid compliance under conditions of high renewable penetration. Machine learning approaches, such as support vector machines, are used to optimize droop parameters for dynamic grid conditions, while deep learning models, including recurrent neural networks, capture complex system dynamics to enhance the stability of distributed energy systems. Reinforcement learning algorithms enable adaptive, autonomous control, improving multi-objective optimization within microgrids. In addition, emerging directions such as transfer learning and real-time data analytics are explored for their potential to enhance scalability and resilience. Overall, this review synthesizes recent advances to demonstrate the growing impact of artificial intelligence in droop control and outlines future pathways toward more intelligent and sustainable power systems. Full article

(This article belongs to the Special Issue Applications of Machine Learning and Artificial Intelligence in Modern Power and Energy Systems, 2nd Edition)

► Show Figures

Graphical abstract

Journal Menu

Journal Browser

Editor’s Choice Articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI