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

Journal Menu

► Journal Menu

Journal Browser

► Journal Browser
share announcement

Editor’s Choice Articles

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

Order results
Result details
Results per page
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
23 pages, 5218 KB  
Article
Light Weight CSI-Based Physical Layer Authentication Model for IoT Networks
by Monika Roopak, Yachao Ran, Simon Parkinson and Jonathon Chambers
Electronics 2026, 15(15), 3350; https://doi.org/10.3390/electronics15153350 - 29 Jul 2026
Viewed by 355
Abstract
This paper introduces a novel physical layer authentication technique for Internet of Things (IoT) networks, leveraging channel state information (CSI) data from Wi-Fi signals to distinguish between authorised and unauthorised nodes and thereby enhancing security without compromising performance. The core novelty lies in [...] Read more.
This paper introduces a novel physical layer authentication technique for Internet of Things (IoT) networks, leveraging channel state information (CSI) data from Wi-Fi signals to distinguish between authorised and unauthorised nodes and thereby enhancing security without compromising performance. The core novelty lies in its integrated framework, which employs non-negative matrix factorization (NMF) for efficient feature selection and a Gaussian mixture model (GMM) for identifying complex patterns within the CSI data specifically adapted to the dynamic nature of IoT networks. NMF is utilised to mitigate the high dimensionality and redundancy inherent in raw CSI metrics, reducing processing load, extracting salient features, alleviating overfitting risks, and exhibiting superior resilience to noise. Following NMF, the GMM component is used for data classification, capitalising on its probabilistic and soft clustering attributes to represent intricate distributions and handle heterogeneous CSI data characteristics. This integrated proposed methodology not only exploits the inherent nonlinear and probabilistic characteristics of CSI data but also upholds computational efficiency, making it highly suitable for implementation in resource-constrained IoT wireless networks. The model achieves exceptional classification proficiency, with an accuracy rate of 99.83 percent and a recall of 100 percent, which are crucial for cybersecurity and anomaly detection. Furthermore, the system is designed for efficiency and minimal resource consumption, exhibiting good computational efficiency, reduced training duration, and lower energy consumption compared with more complex, heavily exploited architectures for CSI data processing like CNN and CNN + LSTM, making it particularly suitable for resource-constrained IoT environments. Full article
(This article belongs to the Special Issue Security and Intelligence for the Next Generation of the Internet of Things)
►▼ Show Figures

Figure 1

18 pages, 7706 KB  
Article
Software-Defined Runtime Reconfiguration of Cryptographic Service Chains for IIoT Edge Nodes
by Lei Zhang, Tianyu Luo, Huiyi Zhao, Chaoen Xiao and Jianxin Wang
Electronics 2026, 15(15), 3325; https://doi.org/10.3390/electronics15153325 - 28 Jul 2026
Viewed by 300
Abstract
In Industrial Internet of Things (IIoT) environments, the security requirements of edge nodes change dynamically, whereas conventional cryptographic deployment relies on static configurations that require firmware upgrades or system reboots for algorithm updates, severely limiting flexibility and maintainability. To address this issue, this [...] Read more.
In Industrial Internet of Things (IIoT) environments, the security requirements of edge nodes change dynamically, whereas conventional cryptographic deployment relies on static configurations that require firmware upgrades or system reboots for algorithm updates, severely limiting flexibility and maintainability. To address this issue, this paper proposes a software-defined runtime reconfiguration mechanism for cryptographic service chains on OS-capable IIoT edge nodes. By decoupling cryptographic processing logic definition from its execution environment, the control plane generates integrity-protected executable service chains, which are dynamically loaded and executed at runtime on the edge node. A prototype implemented on a Raspberry Pi edge node supports dynamic composition and switching among AES, SM4, SHA-256, and SM3, and incorporates HMAC-based integrity verification, version-based rollback prevention, and a smooth transition mechanism supported by kernel buffering. Experimental results show millisecond-level processing latency and a middleware overhead of 3.9–5.2% relative to a static baseline in the same Python interpreter environment, while the reconfiguration latency satisfies the real-time constraints of typical IIoT edge applications. The security and deployment boundaries of the proposed approach are also analyzed. Full article
(This article belongs to the Special Issue New Challenges in IoT Security)
►▼ Show Figures

Figure 1

15 pages, 3150 KB  
Article
Side-Channel Leakage Assessment of an FPGA-Based AES-256 Implementation Using TVLA
by Paweł Rosa and Daniel Waszkiewicz
Electronics 2026, 15(15), 3306; https://doi.org/10.3390/electronics15153306 - 27 Jul 2026
Viewed by 340
Abstract
This paper presents an experimental evaluation of side-channel leakage in an FPGA-based implementation of the AES-256 encryption algorithm operating in ECB mode, with a focus on identifying data-dependent information leakage through power consumption and electromagnetic emissions. The implementation was deployed on a Xilinx [...] Read more.
This paper presents an experimental evaluation of side-channel leakage in an FPGA-based implementation of the AES-256 encryption algorithm operating in ECB mode, with a focus on identifying data-dependent information leakage through power consumption and electromagnetic emissions. The implementation was deployed on a Xilinx Artix-7 FPGA using the ChipWhisperer CW305 platform, while measurements were acquired with the ChipWhisperer Husky system under controlled laboratory conditions. The analysis follows the Test Vector Leakage Assessment methodology in accordance with ISO/IEC 17825, using Welch’s t-test to compare trace sets obtained from fixed and random input data. A dataset of 20,000 traces per configuration was collected, with careful synchronization, interleaving, and preprocessing to ensure statistical reliability. The results show multiple instances where the t-statistic exceeds the threshold of |t| > 4.5 within the defined region of interest, indicating significant leakage. In particular, 84 leakage points were detected in the power consumption channel and 10 in the electromagnetic channel. These findings demonstrate that the evaluated implementation does not satisfy the resistance criteria defined by the standard and remains vulnerable to side-channel analysis, highlighting the need for appropriate countermeasures in FPGA-based cryptographic designs. Full article
(This article belongs to the Special Issue Secure Hardware Architecture and Attack Resilience)
►▼ Show Figures

Figure 1

39 pages, 1370 KB  
Article
TRACE: Feature-Space Feasible Action Contracts for Explainable Intrusion Triage
by Tran Duc Le, Mohammad Arifuzzaman and Yida Bao
Electronics 2026, 15(15), 3291; https://doi.org/10.3390/electronics15153291 - 26 Jul 2026
Viewed by 262
Abstract
Explainable intrusion detection systems often provide feature attributions without indicating whether a security action should be released, downgraded, or deferred. This paper investigates whether action-governed explanations can provide bounded triage evidence under traffic feature feasibility constraints. We present TRACE, a framework that maps [...] Read more.
Explainable intrusion detection systems often provide feature attributions without indicating whether a security action should be released, downgraded, or deferred. This paper investigates whether action-governed explanations can provide bounded triage evidence under traffic feature feasibility constraints. We present TRACE, a framework that maps calibrated detector outputs to a finite action ladder, constructs conformal action sets, selects actions via a utility-minimax rule, and releases high-severity actions only when compact support contracts remain stable under feasible perturbations, where feasibility is a property of the processed benchmark features and not of packet-level realizability. Ablations isolate the conformal set and release gate as the primary drivers of system behavior. Across 11 gated dataset–model pairs, TRACE produces non-degenerate action sets with zero full-set collapse and defer/block rates from 0.603 to 1.000. Under held-out sample split tuning, it achieves higher average proxy utility than unconditional release and release rate-matched random release on all 11 pairs. Against the strongest simple selective gate, however, it matches on 6 of 11 pairs and trails on the remaining 5. Robustness sweeps confirm positive all-row utility on all pairs, though pass-only utility becomes fragile in ultra-low-release regimes. Unlike display-only attribution summaries, the TRACE contract records the plausible action set, feasibility checks, stability summaries, and an explicit release rationale. The results support TRACE as a bounded evidentiary framework for action-governed XAI in IDS, rather than claiming superiority over all IDS/XAI methods or general deployment readiness. 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

27 pages, 3066 KB  
Article
Uncertainty-Aware Machine Learning for Delay-Spread Estimation and Surplus Guard Interval Utilization in IEEE 802.11be Environments
by Jung-Min Moon, Na-Eun Park and Il-Gu Lee
Electronics 2026, 15(15), 3260; https://doi.org/10.3390/electronics15153260 - 24 Jul 2026
Viewed by 366
Abstract
Herein, an uncertainty quantification-based framework is proposed for estimating the root mean square (RMS) delay spread σ as a probability distribution in IEEE 802.11be environments. The method selects a guard interval (GI) using a safety margin derived from the 90th-percentile estimate σ^ [...] Read more.
Herein, an uncertainty quantification-based framework is proposed for estimating the root mean square (RMS) delay spread σ as a probability distribution in IEEE 802.11be environments. The method selects a guard interval (GI) using a safety margin derived from the 90th-percentile estimate σ^q90 and reuses the surplus GI for secondary transmission. The conventional least squares-based estimator exhibits a large ~150 ns RMS error, yielding over 90% underestimation and inefficient GI utilization. To address this, a lightweight one-dimensional convolutional neural network with 28,259 parameters is trained using pinball loss and a softplus monotonicity constraint to produce the q10, q50, and q90 quantiles of the σ distribution. A safety-margin-based policy, GImin = α·σ^q90, is applied, and the surplus GI carries short fast Fourier transform sub-orthogonal frequency division multiplexing secondary transmission. Over IEEE TGn B, D, and E channels across 810,000 trials at α = 3.5, the method reduces RMS error from 150 to 28 ns, lowers the underestimation rate from 92% to 23% regardless of signal-to-noise ratio, and improves goodput by 14% (9.74 to 11.12 Mbps), comparable to the σ-known baseline. The standard-compliant, backward-compatible scheme provides an extensible architecture for next-generation wireless local area networks. Full article
(This article belongs to the Special Issue Innovations in AI and Distributed IoT-Edge-Cloud Computing for Next-Generation Networks)
►▼ Show Figures

Figure 1

18 pages, 828 KB  
Article
Benchmarking and Improving Perceptual Straightening for Robust AI-Generated Video Detection
by Xenofon Papougiannakis and Ruben Tous
Electronics 2026, 15(15), 3245; https://doi.org/10.3390/electronics15153245 - 23 Jul 2026
Viewed by 2519
Abstract
The proliferation of text-to-video generative models—including commercial systems such as Sora, Veo, Runway Gen-3, and Kling—makes automated detection of AI-generated videos an urgent problem in multimedia forensics. We present a study with two interleaved contributions: a cross-family benchmark and two improvements to the [...] Read more.
The proliferation of text-to-video generative models—including commercial systems such as Sora, Veo, Runway Gen-3, and Kling—makes automated detection of AI-generated videos an urgent problem in multimedia forensics. We present a study with two interleaved contributions: a cross-family benchmark and two improvements to the detection pipeline. For the benchmark, we evaluate three methodologically distinct detector families on three public datasets (VidProM, DeepAction, and DeCoF_T2V), spanning diverse open-source and commercial generators: ReStraV, a geometry-supervised detector grounded in the perceptual straightening hypothesis; D3, a training-free detector based on second-order XCLIP temporal statistics; and DeMamba, a supervised Bidirectional Mamba module on frozen XCLIP features. ReStraV and DeMamba achieve broadly comparable global performance on large datasets (VidProM AUC 0.985/0.989; DeCoF_T2V AUC 0.990/0.992), while D3 remains weak as a stand-alone detector under a fixed detection threshold calibrated on VidProM and transferred unchanged to the other benchmarks (AUC 0.534/0.410/0.454 on VidProM/DeepAction/DeCoF_T2V). Per-generator analysis reveals complementary failure modes: DeMamba recovers several hard commercial generators where ReStraV struggles, whereas ReStraV remains competitive in low-data settings at substantially lower computational cost. To bridge these weaknesses, we propose two improvements. First, Rich384 enriches ReStraV’s compact geometric descriptor with DINOv2 temporal embeddings, strengthening ranking quality on large benchmarks and recovering generators that geometry alone misses (AUC 0.995 on VidProM and 0.998 on DeCoF_T2V, with DeepAction AUC decreasing to 0.805). Second, we propose GurAI, our transparent logistic late-fusion method that combines Rich384 and DeMamba logits, raising threshold-free AUC to 0.998/0.999 on VidProM and DeCoF_T2V while preserving interpretability. On the small DeepAction cohort, fusion improves selected per-generator fake recalls at the cost of elevated REAL false positives (AUC 0.769; REAL accuracy 0.460).Our analysis suggests that transparent late fusion can exploit complementary detector strengths more effectively than architectural redesign alone when facing generator diversity. Full article
(This article belongs to the Special Issue Advances in Image Processing and Computer Vision)
►▼ Show Figures

Figure 1

27 pages, 485 KB  
Article
Integrating Side-Channel Power Signals and Network Traffic for Machine Learning-Based Intrusion Detection in IoT
by Felipe Lemus-Prieto, Alejandro Domínguez Campos, José-Luis González-Sánchez and Andrés Caro Lindo
Electronics 2026, 15(14), 3114; https://doi.org/10.3390/electronics15143114 - 15 Jul 2026
Viewed by 380
Abstract
The Internet of Things (IoT) is rapidly being integrated into critical infrastructure sectors, such as energy, transportation, healthcare, and industry. This surge of interconnected devices dramatically expands the attack surface and increases the risk of cascading system failures and data breaches. To address [...] Read more.
The Internet of Things (IoT) is rapidly being integrated into critical infrastructure sectors, such as energy, transportation, healthcare, and industry. This surge of interconnected devices dramatically expands the attack surface and increases the risk of cascading system failures and data breaches. To address these emerging threats, this work proposes an intrusion detection system (IDS) for IoT networks that incorporates machine learning techniques, considering side-channel (power) and network traffic features. We collected power consumption traces and network metrics from IoT devices during normal operation and under diverse cyberattacks (e.g., cryptomining, flooding, port scanning). Time-series machine learning classifiers are trained on this hybrid dataset to differentiate benign versus malicious behavior. The experimental results show that the combined-feature model significantly outperforms models using only one data type, achieving high detection accuracy (F1≈0.89) and correctly identifying the attack type. The resulting IDS generalizes to previously unseen attacks, demonstrating robust, adaptive defense capabilities. The novelty of our approach lies in integrating physical side-channel signals into an automated ML framework, enhancing robustness and resilience. This smart, data-driven solution operates in near real time and helps build autonomous, constantly evolving defenses against cyber threats. Overall, our study delivers a state-of-the-art ML-based tool that learns and evolves to counter modern IoT cyberattacks. Full article
(This article belongs to the Special Issue Emerging Trends in Artificial Intelligence and Intelligent Systems: From Cybersecurity to Smart Societies)
►▼ Show Figures

Figure 1

37 pages, 30432 KB  
Article
Design of an Edge-Cloud IoT System for Dynamic Thermal Sensation Control and Energy Optimization
by Yu Feng Chung, Yu Wen Chu, Yu Ting Kuo and Cheng Ying Chung
Electronics 2026, 15(14), 3088; https://doi.org/10.3390/electronics15143088 - 14 Jul 2026
Viewed by 720
Abstract
Improving HVAC energy efficiency while maintaining collective thermal comfort remains challenging in multi-occupant shared indoor environments, where occupants differ in thermal sensation, activity level, clothing condition, and spatial distribution. This study develops and field-validates an integrated edge-cloud IoT framework that connects non-invasive occupant-state [...] Read more.
Improving HVAC energy efficiency while maintaining collective thermal comfort remains challenging in multi-occupant shared indoor environments, where occupants differ in thermal sensation, activity level, clothing condition, and spatial distribution. This study develops and field-validates an integrated edge-cloud IoT framework that connects non-invasive occupant-state sensing, INT8 edge thermal-sensation inference, and group-comfort-oriented HVAC setpoint optimization for classroom-based shared spaces. The proposed system integrates localized temperature–humidity sensing, vision-derived occupancy, posture, and clothing estimation, cloud-based thermal sensation model training, and edge-deployed real-time control on a HUB 8735 ULTRA device. A 4-day model-training data collection campaign with structured questionnaires was first conducted to obtain occupants’ Thermal Sensation Votes (TSVs) as ground-truth labels. The trained model was compressed from Float32 to INT8 through post-training quantization and deployed on the edge device for real-time inference. Predicted individual TSV values were then transformed into a PPD-inspired TSV-derived dissatisfaction index and used to determine the HVAC setpoint through rolling-horizon group comfort optimization. A separate eight-school-day single-blind daily-block A/B field validation was conducted, with four validation days assigned to the proposed smart control strategy and four days assigned to a fixed 25 °C baseline. The validation dataset included 2194 valid TSV questionnaire responses, which were aggregated into 116 valid 30 min classroom sessions for statistical comparison. The proposed control achieved a session-level mean TSV of −0.13, compared with −0.66 under the baseline, with Welch’s t(100) = 11.2, p < 0.001 and Cohen’s d = 2.11. Daily HVAC energy use decreased from 2.61 to 2.32 kWh/day, corresponding to a cumulative reduction of 1.16 kWh, or 11.1%, over the validation period. These results support the short-term feasibility of the proposed classroom-level human-centric HVAC control framework. However, because the validation was limited to a short-term classroom setting without full weather/load normalization, longer multi-season and multi-room studies are required to further evaluate generalizability and long-term energy performance. Full article
(This article belongs to the Special Issue Advanced Technologies in Signal and Image Processing)
►▼ Show Figures

Figure 1

60 pages, 597 KB  
Review
Heterogeneity-Aware Poisoning Attacks and Mitigation in Federated Learning: A Comprehensive Survey and Taxonomy
by Aimen Djemaa, Djamel Djenouri and Phil Legg
Electronics 2026, 15(13), 2876; https://doi.org/10.3390/electronics15132876 - 1 Jul 2026
Viewed by 635
Abstract
Federated learning (FL) enables collaborative model training without sharing raw data, but remains vulnerable to poisoning attacks in which malicious participants manipulate local data, model updates, gradients, or learned behaviours to degrade performance or introduce targeted failures. These threats become harder to assess [...] Read more.
Federated learning (FL) enables collaborative model training without sharing raw data, but remains vulnerable to poisoning attacks in which malicious participants manipulate local data, model updates, gradients, or learned behaviours to degrade performance or introduce targeted failures. These threats become harder to assess and mitigate in heterogeneous federated learning (HFL), where clients may differ in data distributions, model architectures, task objectives, resource availability, communication reliability, participation patterns, privacy constraints, and deployment environments. Existing surveys provide valuable coverage of FL security, poisoning attacks, robust aggregation, privacy-preserving mechanisms, and heterogeneity, but they do not sufficiently analyse how heterogeneity changes both poisoning behaviour and mitigation reliability. This survey addresses that gap by examining how statistical, model, task, device, communication, and participation heterogeneity affect poisoning feasibility, stealth, persistence, impact, transferability, attribution, and detectability. It then proposes a heterogeneity-aware taxonomy of poisoning mitigation mechanisms and compares existing strategies using operational criteria centred on attack–defence alignment, evidence validity, server visibility, privacy compatibility, scalability, deployment feasibility, and benign-client preservation. The central argument is that poisoning mitigation in HFL should not be evaluated only by attack type, defence family, clean accuracy, or attack success rate but also by whether defences observe and protect the channels through which heterogeneity-shaped attacks are expressed. The survey further identifies open challenges for developing channel-aware, privacy-compatible, scalable, adaptive, and false-positive-aware defences that preserve useful benign diversity under realistic HFL conditions. Full article
(This article belongs to the Special Issue Feature Papers in Networks: 2025–2026 Edition)
►▼ Show Figures

Figure 1

24 pages, 6185 KB  
Article
PILOT: A Replay-Free Continual Learning Approach for Real-Time Semantic Segmentation via Boundary Guidance
by Yujing Zhou, Prashant Shekhar, Thomas Yang and Yongxin Liu
Electronics 2026, 15(13), 2833; https://doi.org/10.3390/electronics15132833 - 29 Jun 2026
Viewed by 351
Abstract
Real-time semantic segmentation models offer an excellent balance between accuracy and inference speed. However, deploying these models in dynamic real-world environments often requires the ability to learn novel classes incrementally without retraining on the entire dataset. This capability is known as continual learning. [...] Read more.
Real-time semantic segmentation models offer an excellent balance between accuracy and inference speed. However, deploying these models in dynamic real-world environments often requires the ability to learn novel classes incrementally without retraining on the entire dataset. This capability is known as continual learning. In this regard, standard fine-tuning methods often suffer from catastrophic forgetting, where the model learns new information but loses accuracy on previously learned classes. The severity of this effect depends on the incremental setup, the available data, and the fine-tuning strategy. Contributing to this crucial domain, this paper proposes a novel continual learning framework tailored for PIDNet, which is a widely cited state-of-the-art real-time semantic segmentation model. Our method, PILOT (Parallel Incremental Learning Over Time), introduces a real-time and lightweight strategy by implementing a parallel Derivative branch (D-branch) designed to capture the high-frequency boundary information of novel classes while freezing the trained parameters of the original segmentation network. This novel setup allows the model to adapt to new semantic categories while preserving the knowledge of previously learned classes. By using only data associated with the new class, our model significantly reduces training overhead. Experimental results demonstrate that our approach successfully segments new classes while maintaining a high mean Intersection over Union (mIoU) on the original base classes, thereby outperforming prior continual learning approaches in this real-time segmentation setting. Overall, PILOT is shown to effectively mitigate catastrophic forgetting with minimal impact on inference latency, adding fewer than 5% additional parameters and reducing the frame rate by only about 9%, thus maintaining real-time performance. Full article
(This article belongs to the Special Issue Cyber-Physical Systems: Recent Developments and Emerging Trends)
►▼ Show Figures

Figure 1

27 pages, 6205 KB  
Article
Low-Latency Machine Vision Based on a Neuromorphic Vision Sensor
by Paul K. J. Park, Junseok Kim, Juhyun Ko and Yeoungjin Chang
Electronics 2026, 15(13), 2828; https://doi.org/10.3390/electronics15132828 - 27 Jun 2026
Cited by 2 | Viewed by 582
Abstract
Low-latency visual perception is essential for interactive machine vision on edge AI devices, but conventional frame-based image sensors impose frame period delays and generate dense image data that increase memory bandwidth and processing latency. Although Dynamic Vision Sensors (DVSs) are known to provide [...] Read more.
Low-latency visual perception is essential for interactive machine vision on edge AI devices, but conventional frame-based image sensors impose frame period delays and generate dense image data that increase memory bandwidth and processing latency. Although Dynamic Vision Sensors (DVSs) are known to provide low latency, sparse output, and high dynamic range, these sensor-level properties do not automatically translate into practical application-level latency reduction on resource-constrained edge platforms. This paper presents a latency-driven sensing algorithm co-design approach for DVS-based low-latency machine vision. The main objective is to connect DVS sensor-level characteristics, event representations, task-dependent processing flows, and measured response times on mobile application processors. We first analyze latency requirements for three representative edge AI applications (i.e., person detection, gesture recognition, and Simultaneous Localization and Mapping (SLAM)), which correspond to different latency regimes and processing structures. We then describe the DVS operating principle, pixel-level event latency, and readout latency, showing how asynchronous event generation reduces sensing delay and suppresses redundant static background information before algorithmic processing. In contrast to prior event camera studies that mainly optimize a single task or a specific event representation, this work evaluates three task-specific event processing systems on mobile processors. Person detection achieves 92 ms processing latency on Exynos 7570, gesture recognition based on event-driven 4-DoF motion estimation achieves 20 ms latency on Exynos 5422, and SLAM achieves 15.9 ms latency on Snapdragon 845. These results satisfy the practical latency targets of the corresponding applications and demonstrate that DVS-based sensing can provide not only sensor-level speed advantages but also system-level latency benefits for AIoT, mobile, robotics, and AR/VR machine vision systems. Full article
(This article belongs to the Special Issue Advances of Artificial Intelligence and Vision Applications: Third Edition)
►▼ Show Figures

Figure 1

29 pages, 3210 KB  
Article
Comparative Evaluation of Gemini and DeepSeek for LLM-Generated Code Quality and Architectural Robustness in Backend Software Engineering
by Marko Horvat, Iva Ursić and Klara Krmpotić
Electronics 2026, 15(13), 2805; https://doi.org/10.3390/electronics15132805 - 25 Jun 2026
Viewed by 2440
Abstract
The increasing integration of large language models (LLMs) into software engineering workflows under the term vibe-coding necessitates systematic empirical evaluation of their code generation capabilities, especially in the context of complex backend development and architectural decision-making. This study compares popular foundational models Google [...] Read more.
The increasing integration of large language models (LLMs) into software engineering workflows under the term vibe-coding necessitates systematic empirical evaluation of their code generation capabilities, especially in the context of complex backend development and architectural decision-making. This study compares popular foundational models Google Gemini 3 Pro and DeepSeek-V3.1 for developing a Java/Spring Boot backend application using a structured prompt-chaining protocol following a typical vibe-coding process. The generated solutions were evaluated using several quantitative and qualitative criteria, including the number of corrective prompts, the extent of required manual code interventions, functional correctness, architectural robustness, maintainability-related design choices, latency, and test quality. The results show substantial differences between the two models. DeepSeek required twice as many corrective natural language prompts as Gemini, but both models required a similar number of manual interventions in the generated code, with 23 for DeepSeek and 20 for Gemini. The most pronounced difference was in architectural reasoning. Gemini autonomously introduced the Data Transfer Object design pattern, resulting in a decoupled architecture, although at a cost of a minor performance issue. In contrast, DeepSeek was better in development of boilerplate code but exposed raw JPA entities through the application interface leading to tight coupling and other issues. Gemini’s solution satisfied 90.25% of evaluated requirements compared to 68.08% for DeepSeek. Additionally, generated tests showed a higher success rate and broader code coverage, achieving 95.7% successful test execution and 55.9% code coverage for Gemini, compared to 74.1% and 45.6% for DeepSeek, respectively. The results indicate that within the paradigm of vibe-coding, even the best available foundational LLMs may still require expert human supervision, especially when the generated code is expected to satisfy specific requirements in production-oriented backend systems. Full article
(This article belongs to the Special Issue AI-Powered Natural Language Processing Applications)
►▼ Show Figures

Figure 1

24 pages, 10388 KB  
Article
Adaptive Content and Style Fusion for Text-to-Image Generations
by Yi-Fang Lee, Chun-Chieh Lee, Chi-Hung Chuang, Chih-Lung Lin and Kuo-Chin Fan
Electronics 2026, 15(13), 2800; https://doi.org/10.3390/electronics15132800 - 25 Jun 2026
Viewed by 466
Abstract
Text-to-image generation aims to produce images that match the semantic content of a text prompt. In style transfer tasks, the model must further integrate reference styles while preserving prompt semantics. However, balancing semantic consistency and style fidelity remains challenging. Existing methods commonly rely [...] Read more.
Text-to-image generation aims to produce images that match the semantic content of a text prompt. In style transfer tasks, the model must further integrate reference styles while preserving prompt semantics. However, balancing semantic consistency and style fidelity remains challenging. Existing methods commonly rely on fixed feature weights and lack adaptive control, which often leads to style over-injection and content distortion. To address these issues, we propose a novel framework that performs dynamic regulation at both the feature and temporal levels. At the feature level, we propose an Entropy-Aware Adaptive Fusion (EAAF) module. It incorporates a bidirectional distribution transformation mechanism to enhance the statistical correlation between content and style features. The module further uses information entropy as a dynamic control signal to adaptively adjust the strength of style injection, thereby achieving a balance between semantic consistency and style fidelity. At the temporal level, we design a Progressive Feature Reweighting (PFR) strategy. By applying stage-wise weighting to content and style features at different diffusion steps, this strategy effectively improves structural stability and color consistency. In addition, our framework is modular and can be integrated into existing diffusion-based style transfer models without additional fine-tuning or retraining. Experimental results demonstrate that applying our approach to current state-of-the-art models, such as StyleStudio and CSGO, significantly enhances their performance, particularly in maintaining strong prompt alignment while achieving high-fidelity style transfer. Full article
(This article belongs to the Special Issue Recent Advances in Object Detection and Computer Vision)
►▼ Show Figures

Figure 1

32 pages, 22420 KB  
Article
FuDensityNet: Occlusion-Aware Multimodal Activation for Robust Object Detection
by Zainab Ouardirhi, Mostapha Zbakh, Mohammed Benjelloun and Sidi Ahmed Mahmoudi
Electronics 2026, 15(13), 2783; https://doi.org/10.3390/electronics15132783 - 24 Jun 2026
Viewed by 385
Abstract
Accurate object detection remains a major challenge in autonomous systems and surveillance, particularly when objects are partially or fully obscured by occlusions. To address this issue, we revisit FuDensityNet as a multimodal detection framework that jointly leverages 2D RGB images and 3D LiDAR [...] Read more.
Accurate object detection remains a major challenge in autonomous systems and surveillance, particularly when objects are partially or fully obscured by occlusions. To address this issue, we revisit FuDensityNet as a multimodal detection framework that jointly leverages 2D RGB images and 3D LiDAR point clouds for robust feature representation. The model integrates spatial and depth cues through low-rank tensor fusion (LRTF) and incorporates an Occlusion Rate (OR) assessment module that estimates the degree of occlusion and dynamically selects the most suitable detection pathway to preserve performance. Experiments on the KITTI and NuScenes datasets indicate that this adaptive strategy improves robustness under high occlusion while maintaining competitive accuracy in less challenging conditions. In particular, FuDensityNet attains 76.6% AP for car detection under “Hard” conditions on KITTI and outperforms several RGB-only and RGB–LiDAR baselines. Owing to its adaptive and modular design, FuDensityNet remains compatible with both 2D and 3D detection pipelines, making it a practical option for real-world environments where visual obstructions are frequent. Full article
(This article belongs to the Special Issue Computer Vision and Machine Learning: Real-World Applications)
►▼ Show Figures

Figure 1

42 pages, 1516 KB  
Review
Agentic AI and Large Language Models for Autonomous IoT Cybersecurity: A Systematic Survey, Taxonomy, and Research Roadmap
by Vinoth Nageshwaran and Soundararajan Ezekiel
Electronics 2026, 15(12), 2740; https://doi.org/10.3390/electronics15122740 - 22 Jun 2026
Cited by 1 | Viewed by 1993
Abstract
Conventional signature-based defenses no longer protect the heterogeneous, large-scale infrastructures that the Internet of Things (IoT) now constitutes. Large language models (LLMs) and agentic artificial intelligence (AI)—systems that autonomously perceive, reason, plan, and act—open a path to self-defending IoT ecosystems, but the integrating [...] Read more.
Conventional signature-based defenses no longer protect the heterogeneous, large-scale infrastructures that the Internet of Things (IoT) now constitutes. Large language models (LLMs) and agentic artificial intelligence (AI)—systems that autonomously perceive, reason, plan, and act—open a path to self-defending IoT ecosystems, but the integrating literature remains fragmented. Within the IEEE Xplore, ACM Digital Library, and MDPI literature, this survey is, to the best of our knowledge, among the first systematic reviews of agentic AI and LLM-driven approaches for autonomous IoT cybersecurity. Following a PRISMA 2020 protocol, we analyze 153 peer-reviewed studies published between 2020 and 2026 in IEEE Xplore, the ACM Digital Library, and MDPI journals. We organize the corpus along a four-pillar taxonomy: agent architecture (single- vs. multi-agent), reasoning strategy (chain-of-thought, ReAct, plan-and-solve, tool use), action scope (detection, response, threat hunting, vulnerability discovery, deception), and deployment topology (edge, fog, cloud). We synthesize four flagship application domains, consolidate datasets and benchmarks, and analyze open challenges including hallucination, prompt-injection robustness, explainability, privacy, latency, and governance. A 2026 research roadmap identifies federated agentic learning, verifiable autonomous reasoning, trustworthy multi-agent collaboration, and resource-hardened edge agents as high-priority directions. A companion reproducibility kit—prompt templates, reference single- and multi-agent loops, and an Edge-IIoTset-style evaluation harness, released as illustrative scaffolding rather than a validated framework—is released publicly and archived on Zenodo (DOI 10.5281/zenodo.20726552). Full article
(This article belongs to the Special Issue AI-Driven Autonomous Cybersecurity Solutions for IoT)
►▼ Show Figures

Figure 1

35 pages, 919 KB  
Article
A Verification-Table-Free Post-Quantum Authenticated Key Agreement Scheme via ML-DSA-Based Subliminal Message Recovery
by Ming-Hsien Lu and Tzung-Her Chen
Electronics 2026, 15(12), 2712; https://doi.org/10.3390/electronics15122712 - 18 Jun 2026
Viewed by 300
Abstract
In user–server authentication environments, persistent server-side verification tables, such as password verifiers, shared authentication records, or per-user secret tables, may become a critical point of failure once leaked. To address this problem in the post-quantum setting, this paper proposes an ML-DSA-specific verification-table-free authenticated [...] Read more.
In user–server authentication environments, persistent server-side verification tables, such as password verifiers, shared authentication records, or per-user secret tables, may become a critical point of failure once leaked. To address this problem in the post-quantum setting, this paper proposes an ML-DSA-specific verification-table-free authenticated key agreement (AKA) scheme based on the NIST-standardized Module-Lattice-Based Digital Signature Algorithm (ML-DSA). The main contribution is a protocol-level use of the signer-recoverable masking vector in ML-DSA as an on-demand reconstruction mechanism for user-related authentication material. This enables the server to reconstruct the required user-related authentication material from its own signature and long-term secret key. This architecture reduces the exposure associated with centralized verification-table leakage, but it should be understood as a storage-relocation tradeoff rather than a storage-free design, because each user must retain the issued signature and the corresponding hash-derived authentication value. By combining the recovered value with identity information through a quantum-resistant one-way hash function, the server can authenticate the user and establish a session key. Its security is analyzed within a Canetti–Krawczyk-style adversarial model and further discussed in the random-oracle setting through a sequence-of-games argument. The analysis supports session-key indistinguishability under the stated freshness and exposure assumptions, while explicitly excluding full forward secrecy under compromise of the server’s long-term ML-DSA secret key. In addition, an operation-level comparison is provided to clarify computational, storage, and communication tradeoffs relative to representative post-quantum AKA schemes. Since the present work does not include implementation-level benchmarking, the performance discussion should be interpreted as analytical rather than empirical validation. The proposed scheme is therefore most suitable for account-login-oriented applications in which reducing centralized verification-table leakage is a primary design objective and where user-side credential storage can be securely managed. Full article
(This article belongs to the Special Issue Digital Security and Privacy Protection: Trends and Applications, 3rd Edition)
►▼ Show Figures

Figure 1

30 pages, 21671 KB  
Article
Semantic Translation and LLM-RAG Fusion of Multi-Source Heterogeneous Data for Production Cognition in Discrete Manufacturing
by Pingwen Zheng, Liping Wang, Changchun Liu and Dunbing Tang
Electronics 2026, 15(12), 2692; https://doi.org/10.3390/electronics15122692 - 17 Jun 2026
Viewed by 403
Abstract
Multi-source heterogeneous data in discrete manufacturing shop floors, including vibration signals, equipment logs, visual monitoring data, and handwritten production reports, exhibit significant differences in modality and semantic representation. Traditional fusion methods often fail to bridge the semantic gap between low-level sensing signals and [...] Read more.
Multi-source heterogeneous data in discrete manufacturing shop floors, including vibration signals, equipment logs, visual monitoring data, and handwritten production reports, exhibit significant differences in modality and semantic representation. Traditional fusion methods often fail to bridge the semantic gap between low-level sensing signals and high-level manufacturing cognition, limiting intelligent anomaly analysis and decision-making capability. To address this issue, this paper proposes a semantic translation and fusion framework for industrial heterogeneous data based on Knowledge Graph (KG), Retrieval-Augmented Generation (RAG), and Large Language Models (LLMs). First, a unified semantic translation mechanism is developed to convert multimodal industrial data into structured semantic representations for cross-modal alignment. Second, an industrial knowledge graph and RAG mechanism are introduced to integrate process knowledge, maintenance manuals, and historical fault records into the reasoning process. Third, an LLM-driven reasoning framework is designed for multimodal semantic fusion, anomaly identification, causal analysis, and optimization recommendation generation. In addition, a digital twin-based visualization interface is constructed to realize real-time interaction between production lines, industrial data, and intelligent cognitive reports. Experimental results demonstrate that the proposed framework significantly improves industrial reasoning accuracy, anomaly analysis correctness, and response efficiency compared with general-purpose LLMs, providing an effective solution for intelligent cognition and decision-making in discrete manufacturing systems. Full article
(This article belongs to the Section Computer Science & Engineering)
►▼ Show Figures

Figure 1

21 pages, 964 KB  
Article
HySV: A Hymba-Inspired Hybrid-Head Framework for Quality-Aware and Deployment-Aware Speaker Verification in Intelligent Embedded Systems
by Sundareswari Thiyagarajan and Deok-Hwan Kim
Electronics 2026, 15(12), 2676; https://doi.org/10.3390/electronics15122676 - 17 Jun 2026
Viewed by 322
Abstract
Speaker verification is an important biometric technology for secure and personalized human–computer interaction in intelligent embedded systems. However, deploying deep speaker verification models on edge devices remains challenging because of restricted computational resources and strict real-time latency requirements. Existing systems commonly rely on [...] Read more.
Speaker verification is an important biometric technology for secure and personalized human–computer interaction in intelligent embedded systems. However, deploying deep speaker verification models on edge devices remains challenging because of restricted computational resources and strict real-time latency requirements. Existing systems commonly rely on convolutional, time-delay, or Transformer-based encoders. Although ECAPA-TDNN-based models provide strong verification performance, their temporal modeling mainly depends on convolutional and TDNN-style operations. Transformer-based models can capture broader temporal patterns, but they often require high computational and memory costs, making them less suitable for embedded deployment. To address these limitations, this paper proposes HySV, a Hymba-inspired hybrid attention and state-space encoder for deployment-aware speaker verification. Rather than directly employing the original Hymba language model, HySV adapts its hybrid-head principle to speaker embedding extraction. Specifically, conventional ECAPA-TDNN-style encoder blocks are replaced with three stacked Hymba context blocks. Each block contains an attention branch for local speaker-discriminative cue modeling and a state-space branch for efficient temporal context summarization. In addition, a quality-aware decision support module is introduced after cosine similarity scoring to improve reliability using utterance duration, voice activity ratio, and embedding confidence. The proposed system is evaluated using both speaker verification and deployment-oriented metrics, including EER, minDCF, FLOPs, and latency. Full article
(This article belongs to the Special Issue Intelligent Embedded Systems: Latest Advances and Applications)
►▼ Show Figures

Figure 1

19 pages, 11225 KB  
Article
Accelerated Graph Neural Networks on an SoC FPGA for Onboard LEO Satellite Network Routing
by Jinhyung Park, Heoncheol Lee, Sungryul Kim, Bongsoo Roh and Myonghun Han
Electronics 2026, 15(12), 2664; https://doi.org/10.3390/electronics15122664 - 16 Jun 2026
Viewed by 521
Abstract
This paper presents a system-on-chip field-programmable gate array (SoC FPGA) acceleration architecture for graph-neural-network- and deep-reinforcement-learning (GNN–DRL)-based routing inference in low-Earth-orbit (LEO) satellite networks. Because LEO satellites move at high orbital speeds, the network topology changes continuously, and routing decisions must track the [...] Read more.
This paper presents a system-on-chip field-programmable gate array (SoC FPGA) acceleration architecture for graph-neural-network- and deep-reinforcement-learning (GNN–DRL)-based routing inference in low-Earth-orbit (LEO) satellite networks. Because LEO satellites move at high orbital speeds, the network topology changes continuously, and routing decisions must track the current link state rather than rely only on static rules. GNN-based DRL routing can represent the graph structure of the network when selecting paths, but its message-passing and readout stages are computationally expensive for resource-constrained onboard platforms. To address this limitation, the trained GNN routing model is ported to an SoC FPGA and implemented with a collaborative processing-system (PS) and programmable-logic (PL) architecture. The PS handles candidate-path generation, environment setup, path selection, and network-state updates, whereas the PL executes the computationally dominant message-passing neural network (MPNN) and readout layers. Post-training INT8 quantization, nonlinear-function approximation, vector-level parallelization, and a parallel multiply–accumulate structure are applied to reduce memory pressure and execution time. Experiments on a ZCU104 board using a PYNQ-controlled PS–PL implementation and an NSFNET-based routing environment show that the proposed PS–PL structure reduces the evaluation time from 94.08 s to 12.63 s compared with the PS-only implementation while maintaining an evaluation score close to that of the original model. Full article
(This article belongs to the Special Issue Recent Advances in AI Hardware Design)
►▼ Show Figures

Figure 1

16 pages, 1369 KB  
Article
A Compact 4T+2T SRAM-Based Digital Compute-in-Memory Bitcell with Reduced Transistor Count for Energy-Efficient Bitwise MAC Operations in 45 nm CMOS
by Shamanth Hariprasad, Srinivas Balasubramanian, Adnan A. Patel and Kyuwon Ken Choi
Electronics 2026, 15(12), 2630; https://doi.org/10.3390/electronics15122630 - 14 Jun 2026
Viewed by 445
Abstract
The increasing computational demands of deep neural network inference drive the need for energy-efficient hardware accelerators that minimize data movement between memory and processing units. Compute-in-memory (CIM) architectures address this bottleneck by embedding computation directly within memory arrays, reducing the overhead of repeated [...] Read more.
The increasing computational demands of deep neural network inference drive the need for energy-efficient hardware accelerators that minimize data movement between memory and processing units. Compute-in-memory (CIM) architectures address this bottleneck by embedding computation directly within memory arrays, reducing the overhead of repeated weight transfers in conventional von Neumann systems. Conventional 6T SRAM-based digital CIM bitcells incur significant transistor overhead as arrays scale, motivating exploration of reduced-transistor bitcell alternatives. We propose a compact 4T+2T SRAM-based digital CIM bitcell implemented in 45 nm CMOS, combining a 4T SRAM storage cell with a 2T multiplier for bitwise multiply-and-accumulate (MAC) operations. The proposed design reduces transistor count from 8 to 6 compared to the 6T+2T reference, lowering parasitic capacitance and hardware overhead without compromising memory or computation functionality. Transient simulations confirm correct write, read, and CIM operations. The bitcell achieves a read delay of 26.91 ps, read power of 1.351 nW, and read energy of 0.005403 fJ—reductions of 98.7%, 86.5%, and 73.1% over the 6T+2T reference, respectively. For CIM operation, bitwise multiplication power decreases from 1.772 µW to 0.8014 µW and energy from 10.63 fJ to 4.808 fJ, representing a 54.8% reduction in both metrics, with only a marginal CIM delay increase of 3.13 ps. Monte Carlo analysis across 100 samples confirms robust write behavior under process variation, with write delay ranging from 55.02 to 69.59 ps and write energy from 0.05870 to 0.06557 fJ. Static noise margin analysis yields an SNM of 83.7 mV under nominal conditions, confirming stable data retention. These results demonstrate that the proposed 4T+2T bitcell offers strong transistor efficiency, energy savings, and computational correctness, making it a promising candidate for area-efficient digital CIM architectures targeting edge AI inference. Full article
(This article belongs to the Special Issue New Challenges in High-Performance Computing and Computer Architecture)
►▼ Show Figures

Figure 1

18 pages, 495 KB  
Article
Beyond Resilience: Antifragility in Critical Infrastructure Cybersecurity
by Stephen Flowerday, Mauricio Papa and Ethan Flowerday
Electronics 2026, 15(12), 2566; https://doi.org/10.3390/electronics15122566 - 10 Jun 2026
Viewed by 408
Abstract
Critical infrastructure cybersecurity increasingly needs frameworks that move beyond recovery toward bounded improvement under disruption, but empirically grounded theories for operational technology remain limited. This paper develops a Theory of Antifragility (AFT) for critical infrastructure (CI) cybersecurity, anchored in a five-state Resilient System [...] Read more.
Critical infrastructure cybersecurity increasingly needs frameworks that move beyond recovery toward bounded improvement under disruption, but empirically grounded theories for operational technology remain limited. This paper develops a Theory of Antifragility (AFT) for critical infrastructure (CI) cybersecurity, anchored in a five-state Resilient System Model and a bounded mathematical definition built around Jensen gain and post-disruption gain. A two-layer empirical design pairs a CI-relevant subset of the CISSM Cyber Events Database with the HAI hardware-in-the-loop industrial control dataset and tests three confirmatory hypotheses and one exploratory proposition. OT-adjacent sectors show significantly higher shares of disruptive or mixed events than comparison sectors (65.3% versus 46.8%, p < 0.001) and a heavier concentration of physical-attack and data-attack subtypes. In HAI, attack-labeled observations were 7.43 times more likely than normal observations to exceed the 95th percentile of baseline deviation (p < 0.001). Across successive attack windows, mean process-state deviation declined significantly (Spearman ρ = −0.688, p = 0.007), providing evidence of measurable response variation rather than proof of adaptive gain. Together, the findings establish the following two prerequisites for future antifragility testing: differentiated fragility burden and process-level perturbation observability. Full article
(This article belongs to the Special Issue Recent Advances in Cybersecurity)
►▼ Show Figures

Figure 1

26 pages, 3678 KB  
Article
LLMSGen: Generating Diverse Seeds for Multimedia Protocol Fuzzing Leveraging Large Language Models
by Chunyang Wan, Yaxin Zhao, Wanli Wu and Kailong Zhu
Electronics 2026, 15(12), 2569; https://doi.org/10.3390/electronics15122569 - 10 Jun 2026
Viewed by 308
Abstract
Multimedia protocol implementations are susceptible to security vulnerabilities, yet the existing fuzzing methods fail to generate seed corpora covering all supported resource types due to the resource type sensitivity characteristic, resulting in limited code coverage. We propose LLMSGen, a method leveraging Large Language [...] Read more.
Multimedia protocol implementations are susceptible to security vulnerabilities, yet the existing fuzzing methods fail to generate seed corpora covering all supported resource types due to the resource type sensitivity characteristic, resulting in limited code coverage. We propose LLMSGen, a method leveraging Large Language Models (LLMs) to generate diverse seeds for multimedia protocol fuzzing. Through prompt chaining, LLMs analyze the source code of the target program to identify supported resource types and extract URI construction rules. Diverse request seeds are then generated by replacing URI fields in captured traffic, and the corresponding resource files are placed in the server’s working directory to ensure valid responses during fuzzing. Evaluated on three RTSP-based open-source programs (Live555, RTSPServer, Gstreamer), LLMSGen increased resource type coverage by an average of approximately 98.4% and improved code line coverage by approximately 36.8% compared to AFLNet’s original seed corpus. Full article
(This article belongs to the Section Networks)
►▼ Show Figures

Figure 1

30 pages, 679 KB  
Article
A Lightweight and Secure End-to-End Authentication Protocol Using PUF for Internet of Drones
by Yeoleum Gang, Hyewon Park and Yohan Park
Electronics 2026, 15(12), 2535; https://doi.org/10.3390/electronics15122535 - 8 Jun 2026
Viewed by 367
Abstract
The Internet of Drones (IoD) has become an important platform for applications such as smart agriculture, industrial monitoring, and large-scale aerial sensing. However, securing IoD communications remains challenging because drones often operate in open environments and have limited computation, storage, and energy resources. [...] Read more.
The Internet of Drones (IoD) has become an important platform for applications such as smart agriculture, industrial monitoring, and large-scale aerial sensing. However, securing IoD communications remains challenging because drones often operate in open environments and have limited computation, storage, and energy resources. Existing authentication and key agreement protocols still face practical limitations, including high computational overhead, exposure to physical capture attacks, and reliance on centralized servers for session-key generation. In this paper, we first analyze a recent IoD authentication scheme and show that it is vulnerable to session-key disclosure, offline identity/password guessing, and mobile device/drone impersonation attacks. To address these issues, we propose a lightweight Physically Unclonable Function (PUF)-based end-to-end authentication protocol for IoD environments. The proposed scheme avoids storing long-term secret keys in drone memory and enables the mobile device and drone to establish a session key directly, without involving the Ground Station Server in key derivation. The security of the proposed protocol is evaluated through informal analysis, BAN logic, the Real-or-Random model, and AVISPA simulation. The results show that the scheme resists common attacks, including replay, impersonation, stolen verifier, physical capture, and offline password guessing attacks. Performance evaluation further indicates that the protocol maintains low computational cost while providing stronger security guarantees, making it suitable for resource-constrained IoD deployments. Full article
(This article belongs to the Special Issue Wireless Sensor Network: Latest Advances and Prospects)
►▼ Show Figures

Figure 1

22 pages, 2560 KB  
Article
An Open Hardware ML-KEM Polynomial Ring Accelerator on Chipyard RISC-V SoC: System-Level Integration and Evaluation
by Yi-Chang Tsai, Yu-Han Lin and Wen-Jyi Hwang
Electronics 2026, 15(12), 2511; https://doi.org/10.3390/electronics15122511 - 7 Jun 2026
Viewed by 615
Abstract
With the standardization of the Module-Lattice-Based Key Encapsulation Mechanism (ML-KEM) in NIST FIPS 203 (2024), efficient hardware support for polynomial ring operations has become critical for practical post-quantum cryptography deployment. The dominant computational workload of ML-KEM arises from matrix–vector multiplications over polynomial rings, [...] Read more.
With the standardization of the Module-Lattice-Based Key Encapsulation Mechanism (ML-KEM) in NIST FIPS 203 (2024), efficient hardware support for polynomial ring operations has become critical for practical post-quantum cryptography deployment. The dominant computational workload of ML-KEM arises from matrix–vector multiplications over polynomial rings, which involve repeated Number Theoretic Transform (NTT), pointwise multiplication, and modular addition operations. This work proposes an ML-KEM polynomial ring accelerator leveraging Open Intellectual Property (Open IP) and integrates it into an open hardware Chipyard RISC-V System on Chip (SoC) via a Memory-Mapped I/O (MMIO) interface. The design incorporates an NTT-based datapath with multiplier and adder arrays, and employs a scratchpad memory to enable intermediate data reuse and reduce memory access overhead. The proposed architecture is implemented on a Genesys 2 FPGA development board featuring a Kintex-7 XC7K325T Field Programmable Gate Array (FPGA) (Digilent Inc., Pullman, WA, USA) and evaluated at both kernel and system levels. Experimental results show that the accelerator reduces matrix–vector multiplication latency to 7372 cycles, achieving up to 40× speedup over a software baseline. At the SoC level, the complete ML-KEM implementation achieves performance improvements of 1.6× to 2.1× across different parameter sets. These results demonstrate that integrating Open IP within an open hardware SoC provides an effective and reproducible approach for accelerating ML-KEM. Full article
(This article belongs to the Special Issue New Trends in Cybersecurity and Hardware Design for IoT)
►▼ Show Figures

Figure 1

39 pages, 1905 KB  
Article
Trust- and Energy-Aware Federated Learning for Wireless Sensor Networks: A Lightweight Orchestration Framework for Heterogeneous IoT Environments
by Manuel J. C. S. Reis, Carlos Serôdio and Frederico Branco
Electronics 2026, 15(11), 2469; https://doi.org/10.3390/electronics15112469 - 4 Jun 2026
Viewed by 277
Abstract
Wireless Sensor Networks (WSNs) are increasingly evolving toward intelligent distributed systems in which local sensing, on-device inference, and collaborative model training are becoming central to scalable Internet of Things (IoT) deployments. However, the practical adoption of Federated Learning (FL) in WSN-oriented environments remains [...] Read more.
Wireless Sensor Networks (WSNs) are increasingly evolving toward intelligent distributed systems in which local sensing, on-device inference, and collaborative model training are becoming central to scalable Internet of Things (IoT) deployments. However, the practical adoption of Federated Learning (FL) in WSN-oriented environments remains constrained by three major challenges: limited and unevenly depleted node energy, heterogeneous non-IID local data distributions, and variable client reliability during collaborative training. This paper proposes a Trust- and Energy-Aware Federated Learning (TEA-FL) framework specifically designed for resource-constrained WSN settings, in which client participation and server-side aggregation are jointly guided by residual energy estimates and dynamically updated trust scores. The proposed method prioritizes reliable, energy-efficient sensor nodes while reducing the impact of weakly aligned or low-quality local updates during global aggregation. The framework is evaluated on two representative WSN/IoT-oriented proxy benchmarks, Human Activity Recognition (HAR) and UNSW-NB15 intrusion detection, under both IID and Dirichlet-based non-IID federated partitions. Under non-IID HAR partitioning, TEA-FL improved final accuracy from 0.6752 with FedAvg to 0.7636 and final Macro-F1 from 0.5623 to 0.7185. On the more challenging non-IID UNSW-NB15 benchmark, TEA-FL achieved the highest final Macro-F1, 0.3711, compared with 0.3230 for FedAvg and 0.3323 for the trust-only baseline, although with a lower final accuracy. These results indicate that TEA-FL is particularly useful when final-round robustness, class-balanced behavior, and client sustainability are more relevant than maximizing a single peak intermediate accuracy value. Additional ablation and unreliable-client experiments further show that the trust–energy-aware aggregation component is particularly influential and that TEA-FL can improve behavior under selected low-quality participation scenarios, although it should not be interpreted as a complete Byzantine-robust defense. Overall, the findings suggest that jointly modeling update consistency and residual energy offers a practical, lightweight pathway toward more dependable and sustainable federated intelligence in next-generation WSN and IoT deployments. Full article
(This article belongs to the Special Issue Wireless Sensor Network: Latest Advances and Prospects)
►▼ Show Figures

Figure 1

21 pages, 2681 KB  
Article
Co-Channel Interference from LEO Satellite Downlinks to 5G-NR Receivers in IMT Spectrum: An Experimental Study
by Massimo Celidonio and Fernando Consalvi
Electronics 2026, 15(11), 2479; https://doi.org/10.3390/electronics15112479 - 4 Jun 2026
Viewed by 557
Abstract
The integration of satellite and terrestrial networks within the same spectrum is a key enabler for extending mobile connectivity in future communication systems. In this context, the Direct Connectivity between Mobile Satellite Service and International Mobile Telecommunications user equipment (DC-MSS-IMT) paradigm, currently under [...] Read more.
The integration of satellite and terrestrial networks within the same spectrum is a key enabler for extending mobile connectivity in future communication systems. In this context, the Direct Connectivity between Mobile Satellite Service and International Mobile Telecommunications user equipment (DC-MSS-IMT) paradigm, currently under study within the International Telecommunication Union foresees the use of terrestrial IMT frequency bands by satellite systems to directly serve conventional mobile devices. This paper presents an experimental study to assess the coexistence between a terrestrial 5G-NR receiver and a co-channel interfering signal representative of a Low Earth Orbit (LEO) satellite downlink. A controlled laboratory setup in a conducted configuration was implemented to ensure repeatability and accurate control of interference conditions. Measurements were performed over four carrier frequencies representative of IMT bands (763 MHz, 1482 MHz, 2150 MHz, and 2635 MHz), considering different traffic load conditions (100% and 50%) and Doppler shifts associated with satellite motion. The interference impact was evaluated in terms of receiver desensitization, defined as the increase in the total received power relative to the baseline noise level. The results show that a 1 dB desensitization threshold is consistently reached when the interfering signal power is approximately 5–6 dB below the receiver noise floor, corresponding to an interference-to-noise ratio (I/N) of about −6 dB. This behavior is observed across all tested frequency bands, traffic conditions, and Doppler scenarios, indicating limited sensitivity to frequency offsets within the considered range. The findings confirm the validity of commonly adopted coexistence criteria and provide experimentally derived reference values to support ongoing regulatory and technical studies on spectrum sharing between satellite and terrestrial IMT systems. Full article
(This article belongs to the Special Issue 5G Non-Terrestrial Networks)
►▼ Show Figures

Figure 1

18 pages, 294 KB  
Article
A Vulnerability Taxonomy for Tor-Based Hidden Services: Toward a De-Anonymization Framework for Cybercrime Investigation
by Jiho Shin and Inkyoung Shin
Electronics 2026, 15(11), 2370; https://doi.org/10.3390/electronics15112370 - 31 May 2026
Viewed by 838
Abstract
Tor-based hidden services host substantial criminal infrastructure, yet de-anonymization research remains fragmented across heterogeneous techniques. No prior work has organized these techniques into a unified taxonomy oriented toward forensic investigation. This paper proposes a five-layer vulnerability taxonomy for Tor hidden services, distinguishing network-level [...] Read more.
Tor-based hidden services host substantial criminal infrastructure, yet de-anonymization research remains fragmented across heterogeneous techniques. No prior work has organized these techniques into a unified taxonomy oriented toward forensic investigation. This paper proposes a five-layer vulnerability taxonomy for Tor hidden services, distinguishing network-level (L1), application-level (L2), side-channel (L3), operational-security-failure (L4), and ecosystem-level (L5) categories. The taxonomy is derived from a structured review of literature published between 2002 and 2024. We further propose a Traceability Evaluation Framework (TEF) that scores 11 vulnerability types along three dimensions: Applicability, Technical Difficulty, and Legal Admissibility. The TEF dimension weights are derived through Analytic Hierarchy Process elicitation from a five-member expert panel of cybercrime investigators, digital forensics researchers, and a legal scholar. The resulting weights of (0.385, 0.204, 0.412) for Applicability, inverted Technical Difficulty, and Legal Admissibility prove robust to ±0.10 perturbations in sensitivity analysis. Under this framework, four application-layer (L2) and operational-security-failure (L4) vulnerabilities receive the highest traceability scores (TS ≥ 2.80), while two network-level (L1) attacks and one side-channel (L3) technique fall to the lowest tier. The framework integrates technical exploitability with legal admissibility constraints across U.S., EU, and other evidentiary regimes, providing a structured reference for investigators and a methodological foundation for case-based empirical validation in future work. Full article
(This article belongs to the Special Issue Advances in Cybersecurity, Data Privacy, Robotics, and Cloud and Service Computing)
18 pages, 18915 KB  
Article
A 140 GHz Two-Channel Transmitter in 40 nm Bulk CMOS
by Junkyu Lee, Changjung Lee, Jaegwan Kim and Munkyo Seo
Electronics 2026, 15(11), 2349; https://doi.org/10.3390/electronics15112349 - 28 May 2026
Viewed by 556
Abstract
This paper presents a 140 GHz two-channel transmitter in 40 nm bulk CMOS technology for D-band wireless communication systems. The transmitter employs a direct upconversion architecture with IQ Gilbert cell mixers and a shared ×9 frequency multiplier for local oscillator (LO) generation. [...] Read more.
This paper presents a 140 GHz two-channel transmitter in 40 nm bulk CMOS technology for D-band wireless communication systems. The transmitter employs a direct upconversion architecture with IQ Gilbert cell mixers and a shared ×9 frequency multiplier for local oscillator (LO) generation. The Lange coupler generates quadrature LO signals for I and Q paths, while the two-way four-stage differential power amplifier with cascade topology provides high output power. On-wafer measurement at 140 GHz LO frequency demonstrates a 9.9 dB conversion gain with a 5.5–6.1 GHz 3 dB bandwidth. The measured saturated output power is 10.1 dBm with an output 1 dB compression point of 6.5 dBm. The IQ imbalance remains within 2 dB across the 3 dB bandwidth. The fabricated transmitter occupies a chip area of 1.68 mm2 and consumes 435 mW from a 1 V supply. The power density of 6.09 mW/mm2 is the highest among reported CMOS-based D-band transmitters. The dual-channel architecture with shared LO generation enables MIMO transmission, spatial multiplexing, and diversity techniques while maintaining compact size and competitive power efficiency for high data rate wireless applications in the D-band frequency range. Full article
(This article belongs to the Section Microwave and Wireless Communications)
►▼ Show Figures

Figure 1

24 pages, 2369 KB  
Article
A Single-Link Propagation-Driven Performance Study of IEEE 802.11be Wi-Fi 7 in Complex Indoor Environments
by Nurul I. Sarkar and Rashid Mustafa
Electronics 2026, 15(11), 2324; https://doi.org/10.3390/electronics15112324 - 27 May 2026
Cited by 1 | Viewed by 814
Abstract
IEEE 802.11be, commercially known as Wi-Fi 7, extends wireless local area network (WLAN) capability through wider channel bandwidths, higher-order modulation, and tri-band operation. However, realised indoor performance is still strongly affected by radio propagation conditions. This study presents a controlled empirical assessment of [...] Read more.
IEEE 802.11be, commercially known as Wi-Fi 7, extends wireless local area network (WLAN) capability through wider channel bandwidths, higher-order modulation, and tri-band operation. However, realised indoor performance is still strongly affected by radio propagation conditions. This study presents a controlled empirical assessment of Wi-Fi 7 behaviour in a multi-storey university building by examining throughput and received signal strength (RSS) across the 2.4 GHz, 5 GHz, and 6 GHz bands using a single-link measurement setup. Six experimental scenarios were used to examine distance variation, wall penetration, line-of-sight (LOS) obstruction, floor separation, antenna orientation, and microwave interference. The measured RSS values were compared with the free-space, two-ray ground reflection, and log-distance shadowing models using mean absolute error (MAE). Six experimental scenarios were designed to isolate dominant indoor impairments, including distance variation, wall penetration, line-of-sight obstruction, floor separation, antenna orientation, and microwave interference. Measured RSS values were evaluated against free-space, two-ray, and log-distance shadowing models using mean absolute error as the comparison metric. Results show that 2.4 GHz retains greater penetration at lesser capacity, while 6 GHz offers the maximum short-range throughput under clear line-of-sight conditionsbut rapidly deteriorates with structural attenuation. Performance in all bands is greatly diminished by multi-wall blockage and line-of-sight loss. A single propagation model cannot adequately capture the divergence introduced by increasing distance and indoor attenuation, while short-range line-of-sight conditions more closely resemble deterministic predictions in terms of measured RSS alignment. Overall, the results highlight the trade-off between Wi-Fi 7’s capacity and coverage, and provide helpful advice for choosing frequencies, positioning access points, and organizing indoor coverage. The research findings provide insights into the practical deployment of next-generation Wi-Fi in multi-story buildings and residential houses. Full article
(This article belongs to the Special Issue Wireless Artificial Intelligent Computing Systems and Applications (WASA))
►▼ Show Figures

Figure 1

31 pages, 13410 KB  
Article
Early Detection of Distributed Denial of Service in Cloud Computing Using Quantum-Enhanced Knowledge Distillation Framework
by Bhargavi Krishnamurthy, Saikat Das and Sajjan G. Shiva
Electronics 2026, 15(11), 2327; https://doi.org/10.3390/electronics15112327 - 27 May 2026
Viewed by 371
Abstract
Cloud computing is one of the essential computing platforms for modern enterprises. About 98 percent of large businesses will use cloud computing services in 2025 to enable remote working. The highly distributed structures of cloud computing are prone to attacks starting from weakened [...] Read more.
Cloud computing is one of the essential computing platforms for modern enterprises. About 98 percent of large businesses will use cloud computing services in 2025 to enable remote working. The highly distributed structures of cloud computing are prone to attacks starting from weakened access control to data breaches. The sources making cloud systems vulnerable to attacks are public accessibility, auto scaling, and shared form of network architecture. Distributed Denial of Service (DDoS) is one of the most serious forms of attacks where multiple botnets get created simultaneously and flood massive requests for the cloud services. If the DDoS attack is not identified early it leads to the unavailability of cloud services, increased cost of migration, exhaustion of resources, and frequent violations of Service Level Agreements (SLAs). Hence, there is a need to detect DDoS at an early stage. Traditional machine learning models demand high computational power and larger memory capacity which make it unsuitable for a real-time cloud environment. This limitation is overcome by presenting a novel Quantum-Enhanced Knowledge Distillation framework (QKD) to detect DDoS attacks in cloud systems. QKD is a highly potential form of architecture which uses quantum computing to enhance the knowledge transfer between teacher and student models. The knowledge is extracted from the teacher model and quantum encoding of knowledge is performed. The complex correlation between the features of the traffic is extracted by applying the entanglement gates. The student model is trained considering the distillation loss and optimized until convergence. The simulation of the QKD is performed using DynamicCloudSim 3.0.3 simulator considering benchmark dataset CIC-DDoS2019and the performance is further validated using expected value analysis methodology. The performance of QKD is found to be promising toward performance metrics such as packet loss rate, attack detection time, attack recovery ratio, bandwidth utilization, and response time. Full article
(This article belongs to the Special Issue Artificial Intelligence/Machine Learning (AI/ML)-Based Tools for Cyber Security)
►▼ Show Figures

Figure 1

32 pages, 3646 KB  
Article
Client-Side Continuous Authentication Using Keystroke Dynamics: A Lightweight Pipeline and Cross-Session Evaluation
by Zhanhe Zhang, Maria Papaioannou, Gaurav Choudhary and Nicola Dragoni
Electronics 2026, 15(11), 2325; https://doi.org/10.3390/electronics15112325 - 27 May 2026
Viewed by 543
Abstract
Post-login threats such as device sharing and session takeover motivate continuous authentication with behavioral signals. This paper studies a lightweight keystroke-dynamics pipeline designed for strict cross-session evaluation and browser-side scoring. Using the fixed-text and free-text tracks of the public KeyRecs dataset, we extract [...] Read more.
Post-login threats such as device sharing and session takeover motivate continuous authentication with behavioral signals. This paper studies a lightweight keystroke-dynamics pipeline designed for strict cross-session evaluation and browser-side scoring. Using the fixed-text and free-text tracks of the public KeyRecs dataset, we extract compact repetition-level and sliding-window digraph-timing features and train per-user one-vs-rest Logistic Regression verifiers on Session 1 (S1). Thresholds are selected only on S1 and transferred unchanged to Session 2 (S2), preventing test-set tuning and exposing operating-point instability under session drift. Fixed-text achieves S2 AUC mean/median 0.895/0.918 with a half total error rate (HTER) around 0.19, while free-text reaches AUC mean/median 0.884/0.899 with a similar transferred-threshold HTER. Personal thresholds and a pooled-S1 global threshold perform similarly on average, suggesting that global thresholding can simplify deployment without replacing per-user scoring models. A scaler-only warm-up update yields limited and inconsistent gains, showing that mean/variance adaptation alone is insufficient. Finally, compact JSON artifacts and replay-based browser benchmarks demonstrate deterministic client-side scoring with very small per-sample latency. Overall, the results show that useful threshold-free separability does not by itself guarantee stable operating-point transfer under cross-session drift. Full article
(This article belongs to the Special Issue Improving IoT Security and Efficiency Through Advanced Data Analysis Method)
►▼ Show Figures

Figure 1

37 pages, 2001 KB  
Article
Spec2SeqFuzz: A Category Prediction-Guided Approach for Stateful Multi-Step REST API Fuzzing
by Zhuofeng He, Sunpei Shang, Yumeng Guo and Aojie Zhou
Electronics 2026, 15(11), 2309; https://doi.org/10.3390/electronics15112309 - 26 May 2026
Viewed by 443
Abstract
REST APIs have become a dominant interface for modern web applications and cloud services, and a growing body of work has studied automated testing and reproducible error discovery for such systems. Prior approaches have explored dependency inference, cross-request value reuse, and, more recently, [...] Read more.
REST APIs have become a dominant interface for modern web applications and cloud services, and a growing body of work has studied automated testing and reproducible error discovery for such systems. Prior approaches have explored dependency inference, cross-request value reuse, and, more recently, learning- or LLM-based test generation. However, deep stateful multi-step reproducible error discovery remains difficult in practice because sequence construction is still often performed directly in the endpoint space, reusable runtime artifacts are not always tightly coupled with sequence expansion, and online LLM-driven generation may introduce cost and instability. We present Spec2SeqFuzz, a stateful multi-step fuzzing framework for REST API systems. The central idea is to guide online exploration in a compact category space rather than directly in the full endpoint space. Spec2SeqFuzz uses LLMs only in an offline pre-processing stage to normalize public multi-step PoCs, classify OpenAPI endpoints into a transferable category taxonomy, and construct training data for next-category prediction. During online fuzzing, the framework predicts the next likely API category from the executed prefix and observed response feedback, maps the predicted categories back to concrete endpoints, and combines this guidance with black-box endpoint fuzzing, proxy-based payload collection, and snapshot-assisted state restoration. We implemented a prototype and evaluated it on GitLab and WordPress, using MINER as the primary reproduced baseline in our current study. The results show that Spec2SeqFuzz is promising for both multi-step and single-endpoint error discovery on these two targets. Following the terminology used in MINER, we report reproducible errors rather than treating every triggered failure as a confirmed security vulnerability. Across the two targets, Spec2SeqFuzz discovers more reproducible multi-step errors than MINER, while the ablation results further suggest that category guidance, payload reuse, and depth-first stateful exploration are important to the final error-discovery performance. Full article
(This article belongs to the Special Issue Advances in Cyber-Physical Networks: Intelligence, Control and Security)
►▼ Show Figures

Figure 1

31 pages, 917 KB  
Article
X-GATE: Attribution-Aware Distillation and Hardening for Compressed Edge-IIoT Intrusion Detection
by Tran Duc Le, Yida Bao and Mohammad Arifuzzaman
Electronics 2026, 15(11), 2284; https://doi.org/10.3390/electronics15112284 - 25 May 2026
Viewed by 409
Abstract
Industrial Internet of Things (IIoT) intrusion detection requires compact, latency-efficient models whose behavior remains assessable under adversarial stress, yet compression can alter the feature-attribution structure learned by a full-precision model. This paper presents X-GATE (eXplanation-Guided Adversarial Training Engine), an attribution-aware training framework for [...] Read more.
Industrial Internet of Things (IIoT) intrusion detection requires compact, latency-efficient models whose behavior remains assessable under adversarial stress, yet compression can alter the feature-attribution structure learned by a full-precision model. This paper presents X-GATE (eXplanation-Guided Adversarial Training Engine), an attribution-aware training framework for compressed Edge-IIoT intrusion detection. X-GATE combines Explanation-Consistency Distillation (ECD), which aligns Teacher–Student feature-attribution rankings with a differentiable soft-rank Spearman penalty, and Explanation-Guided Adversarial Training (EGAT), which hardens the Student on Teacher-salient feature coordinates. On the full Edge-IIoTset 2022 benchmark, the latest three-seed ablation gives Full X-GATE 89.30 ± 3.89% F1-Macro with 0.617 M parameters, within approximately 0.6 percentage points of the full-precision Teacher; a Random Forest model remains a stronger clean-F1 reference, so X-GATE is not framed as the clean-accuracy optimum. In a separate deployment-subset rerun, X-GATE obtains 78.83 ± 5.83% float F1-Macro and 79.11 ± 5.47% INT8 F1-Macro, reduces the adversarial false-positive rate from 0.46 ± 0.08% for KD-only to 0.16 ± 0.09% under the evaluated single-step white-box explanation-evasion protocol, and reduces CPU latency from 4.16 to 1.25 ms/sample. Component ablation further shows that ECD reduces Logical Drift by 17.24%, while EGAT improves adversarial F1 by 10.57 percentage points. Taken together, these benchmark- and protocol-bounded results position X-GATE as a compact neural operating point for the Edge-IIoT setting studied here, balancing attribution consistency, targeted hardening, and CPU-side efficiency. 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

16 pages, 412 KB  
Article
Exploring the Effects of Data Volume and Transfer-Language Choice on Transfer Learning with Application to Polish
by Juuso Eronen, Zhenzhen Liu, Michal Ptaszynski, Karol Nowakowski and Fumito Masui
Electronics 2026, 15(11), 2254; https://doi.org/10.3390/electronics15112254 - 22 May 2026
Viewed by 371
Abstract
Transfer learning offers a practical way to improve neural machine translation in low-resource settings, but its effectiveness depends on both the choice of transfer language and the amount of target-language data available for adaptation. In this study, we examine these factors specifically for [...] Read more.
Transfer learning offers a practical way to improve neural machine translation in low-resource settings, but its effectiveness depends on both the choice of transfer language and the amount of target-language data available for adaptation. In this study, we examine these factors specifically for Polish–English translation using mBART. We evaluate Czech, Russian, and German as parent languages and extend the analysis with a combined Slavic parent model trained on Czech and Russian. The models are compared across 0-shot, 10-shot, 100-shot, 1k-shot, and 10k-shot settings. Within this Polish–English mBART setting, Czech provides the strongest zero-shot performance, while Russian and German improve substantially as Polish fine-tuning data increases and achieve the strongest results at higher shot levels. The paper therefore analyzes selected transfer-language configurations rather than a formally measured similarity variable. The results suggest that, in this setup, transfer-language choice matters most when no Polish supervision is available, whereas larger amounts of Polish data can compensate for weaker initial transfer alignment. Full article
(This article belongs to the Special Issue Natural Language Processing: From Deep Learning to Real-World Solutions)
►▼ Show Figures

Figure 1

19 pages, 2536 KB  
Article
A Lightweight Network for Encrypted Traffic Classification Based on Convolutional Positional Encoding and Efficient Multi-Scale Attention
by Yuan Feng, Yifan Ren, Jianwei Zhang, Zengyu Cai, Juncheng Yang and Liang Zhu
Electronics 2026, 15(11), 2248; https://doi.org/10.3390/electronics15112248 - 22 May 2026
Viewed by 487
Abstract
Network traffic classification is a cornerstone of network management and security. Addressing the challenges of feature extraction in encrypted traffic and the deployment limitations of traditional deep learning models on resource-constrained edge devices due to their large parameter sizes, this paper proposes a [...] Read more.
Network traffic classification is a cornerstone of network management and security. Addressing the challenges of feature extraction in encrypted traffic and the deployment limitations of traditional deep learning models on resource-constrained edge devices due to their large parameter sizes, this paper proposes a lightweight network for encrypted traffic classification, termed CEMA-Net (Convolutional Positional Encoding and Efficient Multi-scale Attention Network). Specifically, the proposed model integrates an Efficient Multi-scale Attention (EMA) mechanism with a Convolutional Positional Encoding (CPE) strategy to jointly capture global dependencies and local contextual information. To enable efficient adaptation to traffic data, an Efficient Multi-scale Attention Adapter (EMAAdapter) is designed, which reconstructs one-dimensional traffic sequences into a pseudo-2D representation and extracts horizontal, vertical, and local features in parallel. This design facilitates effective modeling of complex cross-scale dependencies in encrypted traffic with minimal computational overhead. Experimental results on three public datasets demonstrate that the proposed method, with only 0.66 M parameters, achieves superior classification performance compared with mainstream vision-based models such as ResNet-101, while significantly reducing computational cost. These results highlight the effectiveness of combining convolutional positional encoding with multi-scale attention mechanisms and provide an efficient solution for encrypted traffic classification in resource-constrained environments. Full article
(This article belongs to the Section Networks)
►▼ Show Figures

Figure 1

26 pages, 2578 KB  
Article
Ontological Representation of Cyber–Physical Systems for Knowledge-Based Production
by Kathrin Gorgs, Tom Löhnert, Tobias Vogel and Matthias L. Hemmje
Electronics 2026, 15(11), 2235; https://doi.org/10.3390/electronics15112235 - 22 May 2026
Viewed by 452
Abstract
This paper presents a process-centric ontology for the semantic representation of cyber–physical systems (CPSs) within knowledge-based production planning (KPP). The approach integrates physical systems (PSs), cyber systems (CSs), and CPSs into a unified semantic model based on a three-layer classification. The ontology was [...] Read more.
This paper presents a process-centric ontology for the semantic representation of cyber–physical systems (CPSs) within knowledge-based production planning (KPP). The approach integrates physical systems (PSs), cyber systems (CSs), and CPSs into a unified semantic model based on a three-layer classification. The ontology was implemented using OWL and integrated into a Neo4j-based graph architecture to support semantic querying and process modeling. The evaluation was conducted using prototypical manufacturing scenarios, including semiconductor and mechanical engineering domains. Validation included (i) consistency checking using the HermiT reasoner, (ii) execution of SPARQL queries for retrieving CPS-related process information, and (iii) integration into a three-stage planning model. The results show that the ontology enables consistent semantic representation and cross-domain querying of CPS-based production processes. The work provides a validated proof-of-concept and establishes a foundation for future research on ontology-based production systems. Full article
(This article belongs to the Section Computer Science & Engineering)
►▼ Show Figures

Figure 1

37 pages, 2822 KB  
Article
A Real-Time Sensor-Driven Multi-Agent Navigation System with Reinforcement Learning for Blind and Visually Impaired Users in Urban Environments
by Pilar Herrero-Martin and Álvaro García-Ballestero
Electronics 2026, 15(11), 2250; https://doi.org/10.3390/electronics15112250 - 22 May 2026
Viewed by 575
Abstract
Urban navigation in dynamic environments remains a challenging problem for blind and visually impaired users due to the presence of unpredictable obstacles and the limitations of conventional navigation systems, which rely primarily on static map-based information and lack real-time environmental awareness. This paper [...] Read more.
Urban navigation in dynamic environments remains a challenging problem for blind and visually impaired users due to the presence of unpredictable obstacles and the limitations of conventional navigation systems, which rely primarily on static map-based information and lack real-time environmental awareness. This paper presents a real-time sensor-driven navigation system based on a multi-agent architecture incorporating a reinforcement-learning navigation policy for assistive mobility in urban environments. The proposed system integrates GPS-based global localization with vision-based perception to enable continuous fusion of global route planning and local obstacle detection. This integration allows the system to dynamically adjust navigation strategies in response to changing environmental conditions. The architecture is designed as a modular multi-agent system comprising agents for perception, navigation, sensor fusion, personalization, safety arbitration, interface management, and system monitoring. The reinforcement learning component formulates local navigation as a sequential decision-making problem, where the navigation policy is trained to balance path efficiency, obstacle avoidance, and safety constraints through interaction with simulated environments. Prototype implementation is developed and evaluated in both simulation and controlled real-world scenarios. Experimental results demonstrate that the proposed system shows improved obstacle avoidance performance and navigation stability under the evaluated conditions while maintaining low-latency responsiveness compared to baseline navigation approaches. The system also exhibits robust behaviour under varying environmental conditions, supporting its potential applicability to assistive navigation tasks in controlled urban environments. The proposed approach contributes to a scalable architecture that integrates a reinforcement-learning navigation policy within a multi-agent coordination framework and real-time sensor perception, providing a foundation for the development of intelligent and deployable assistive navigation systems. Full article
(This article belongs to the Special Issue Wearable and Intelligent (Bio)Electronics for Healthcare: From Sensors to AI)
►▼ Show Figures

Figure 1

19 pages, 692 KB  
Article
A Hierarchical Key Management Scheme for Efficient Outsourced Computation in Cloud Storage Services
by Nan-I Wu, Tsuei-Hung Sun, Cheng-Ying Yang and Min-Shiang Hwang
Electronics 2026, 15(10), 2185; https://doi.org/10.3390/electronics15102185 - 19 May 2026
Viewed by 386
Abstract
This study addresses the high computational overhead associated with updating encryption keys and authentication information by delegating these tasks to cloud storage services. This approach reduces the burden on data owners and facilitates the establishment of a robust cloud storage ecosystem. Our evaluation [...] Read more.
This study addresses the high computational overhead associated with updating encryption keys and authentication information by delegating these tasks to cloud storage services. This approach reduces the burden on data owners and facilitates the establishment of a robust cloud storage ecosystem. Our evaluation focuses on three critical performance metrics for cloud security mechanisms: 1. Computational Complexity: The processing time and frequency required for encryption, decryption, and verification. 2. Communication Complexity: The volume and length of message exchanges. 3. Storage Overhead: The space required to maintain a comprehensive cloud storage system. We propose a hierarchical key management scheme that enables data owners to update file encryption keys efficiently without impacting other users within the Cloud Storage Service (CSS) environment. By eliminating the need for owners to store all keys in a hierarchical tree, our scheme minimizes the computational cost of key updates. Ultimately, this approach enhances efficiency for both data owners and receivers, making it particularly suitable for resource-constrained mobile devices. Full article
(This article belongs to the Special Issue Digital Security and Privacy Protection: Trends and Applications, 3rd Edition)
►▼ Show Figures

Figure 1

18 pages, 429 KB  
Article
Evaluating Distributed Communication Architectures for GPU-Accelerated Image Encoding
by Haojie Zheng, Carlos Reaño and Juan F. Ariño-Sales
Electronics 2026, 15(10), 2137; https://doi.org/10.3390/electronics15102137 - 16 May 2026
Cited by 1 | Viewed by 481
Abstract
Artificial intelligence (AI) has transformed how we engage with visual data, particularly within the context of enterprises. Multi-modal codification systems enable the creation of semantic connections between text and visual data using AI models. This opens new markets for businesses by enabling visual [...] Read more.
Artificial intelligence (AI) has transformed how we engage with visual data, particularly within the context of enterprises. Multi-modal codification systems enable the creation of semantic connections between text and visual data using AI models. This opens new markets for businesses by enabling visual search engines, recommendation systems, and automatic tagging of visual data. However, implementing these systems presents significant technical challenges. The typical workflow involves encoding images using an AI model, converting these representations into semantic vectors, and inserting them into databases optimized for fast searches. This not only affects technical efficiency but also impacts the ability of companies to scale these systems to a commercial level. This paper presents a comprehensive comparative analysis of communication architectures for large-scale image encoding systems, evaluating gRPC, RabbitMQ, serverless Lambda, and SageMaker approaches across performance and resource efficiency dimensions. Through controlled experiments processing up to 18,000 images using the SigLIP model, we establish clear performance–architecture relationships that inform system design decisions for visual content-based search applications. Full article
(This article belongs to the Special Issue AI-Based Image Processing Detection and Classification Analysis for Multidisciplinary Approaches)
►▼ Show Figures

Figure 1

29 pages, 2292 KB  
Article
EcoInfer: Optimizing Energy Efficiency with Latency Guarantees Through Iteration-Level GPU Frequency Control in LLM Serving
by Qingyuan Hu and Jian Li
Electronics 2026, 15(10), 2139; https://doi.org/10.3390/electronics15102139 - 16 May 2026
Viewed by 815
Abstract
Large language model (LLM) serving has emerged as a major source of energy consumption in modern AI infrastructure. In current deployments, graphics processing units (GPUs) are typically operated at default high-frequency settings to maximize performance. However, under practical service-level objectives (SLOs), peak performance [...] Read more.
Large language model (LLM) serving has emerged as a major source of energy consumption in modern AI infrastructure. In current deployments, graphics processing units (GPUs) are typically operated at default high-frequency settings to maximize performance. However, under practical service-level objectives (SLOs), peak performance is often unnecessary, especially during the memory-bound decode stage, resulting in substantial power redundancy and avoidable energy waste. Existing studies that apply GPU dynamic voltage and frequency scaling (DVFS) to improve the energy efficiency of LLM serving have shown promising results. However, they generally rely on coarse-grained control, accurate output length prediction, or request-level resource management, which limits their effectiveness under highly dynamic workloads and strict SLO constraints. We present EcoInfer, a fine-grained DVFS framework for energy-efficient LLM serving. EcoInfer performs iteration-level, workload-aware GPU frequency control that adapts to the current inference phase and system state while preserving latency guarantees. It comprises three tightly integrated modules: a machine-learning-based frequency–latency predictor that estimates iteration latency across candidate GPU frequencies using lightweight iteration-level features; an SLO-aware frequency controller that selects the minimum feasible frequency within a sweet-spot-guided candidate range; and a low-overhead runtime optimization layer that combines adaptive decision caching with asynchronous execution to reduce and hide the overhead of online control. Implemented on top of vLLM, EcoInfer achieves up to 25.4% energy savings and 21.5% average energy savings and improves energy efficiency by 1.28× on average in terms of Tokens/J while maintaining a nearly unchanged SLO attainment rate compared with the default vLLM baseline. Full article
(This article belongs to the Special Issue High-Performance Computing for AI: Architecture, Systems, and Algorithms)
►▼ Show Figures

Figure 1

30 pages, 2091 KB  
Article
MOSAIC: A Cognitively Motivated Multi-Agent Framework for Interpretable and Training-Free Empathetic Dialogue
by Kai Liu, Hangyu Xiong, Jinyi Zhang and Min Peng
Electronics 2026, 15(10), 2078; https://doi.org/10.3390/electronics15102078 - 13 May 2026
Viewed by 486
Abstract
Empathetic dialogue systems built upon large language models overwhelmingly adopt a monolithic inference paradigm that processes emotion perception, causal reasoning, memory retrieval, and response planning within a single forward pass without architecturally enforced intermediate representations, forfeiting intermediate-state transparency and long-horizon personalization. Drawing on [...] Read more.
Empathetic dialogue systems built upon large language models overwhelmingly adopt a monolithic inference paradigm that processes emotion perception, causal reasoning, memory retrieval, and response planning within a single forward pass without architecturally enforced intermediate representations, forfeiting intermediate-state transparency and long-horizon personalization. Drawing on neuroscientific and cognitive–psychological evidence that human empathy is functionally dissociable, we present MOSAIC (Multi-agent Orchestration with Structured Affective memory for Interpretable empathiC dialogue), a training-free framework that operationalizes empathetic dialogue as a four-stage cognitive pipeline: affective perception, causal appraisal, episodic memory retrieval, and response synthesis. Three innovations distinguish MOSAIC from prior work: (1) a cognitively motivated modular architecture whose functionally dissociable stages enable post hoc failure attribution through logged intermediate states; (2) a hierarchical three-tier emotional memory—perceptual, semantic, and episodic—coupled with adaptive three-dimensional retrieval over emotion, situation, and coping-strategy cues; and (3) a heterogeneous model orchestration strategy coordinating open-source and API-accessible models through role-specific chain-of-thought prompts, requiring no task-specific fine-tuning. We note that the EmpatheticDialogues evaluation pre-populates the memory store with 200 training-split episodes prior to test-set interaction, a data-access asymmetry relative to single-model baselines that must be borne in mind when interpreting comparative results. Experiments on EmpatheticDialogues and ESConv show that MOSAIC achieves a 76.4% weighted F1 and an empathy score of 3.87 (on a 1–5 Likert scale) and that it improves over single-model, training-free baselines on aggregate empathy and—most prominently—on human-rated personalization (3.67 vs. 3.24 against Claude-3.5 five-shot, d=0.48). We caution that the comparison against training-free baselines is not data access-controlled (see the cold-start discussion in Methods); the personalization advantage, supported by the ablation without the Event Agent, is the result we treat as the primary practical contribution of this work. Full article
(This article belongs to the Special Issue Affective Computing in Human–Robot Interaction)
►▼ Show Figures

Figure 1

16 pages, 26838 KB  
Article
Experimental Evaluation and Theoretical Analysis of I/Q Imbalance in Direct Millimeter-Wave Six-Port QPSK Demodulators
by Chaouki Hannachi, Matthieu Egels, Phillipe Pannier and Serioja Ovidiu Tatu
Electronics 2026, 15(10), 2072; https://doi.org/10.3390/electronics15102072 - 13 May 2026
Viewed by 430
Abstract
This paper presents a comprehensive investigation of the impact of I/Q (In-phase/Quadrature) imbalance on the performance of a six-port receiver operating in the millimeter-wave band, specifically in the 60–65 GHz frequency range. Unlike traditional heterodyne architectures, the six-port junction offers a low-cost and [...] Read more.
This paper presents a comprehensive investigation of the impact of I/Q (In-phase/Quadrature) imbalance on the performance of a six-port receiver operating in the millimeter-wave band, specifically in the 60–65 GHz frequency range. Unlike traditional heterodyne architectures, the six-port junction offers a low-cost and low-power alternative for direct conversion; however, it is highly sensitive to hardware imperfections. This study demonstrates that manufacturing tolerances in passive components, such as 90° hybrid couplers and power dividers, introduce significant amplitude and phase disparities. These imbalances geometrically distort the ideal QPSK constellation, transforming the circular decision boundaries into an elliptical profile. The research methodology employs a robust co-simulation approach in Advanced Design System (ADS), integrating measured S-parameters with mathematical analysis to quantify signal degradation. Performance is evaluated using the Error Vector Magnitude (EVM) metric. The experimental findings reveal that even at the higher end of the spectrum (65 GHz), where the amplitude imbalance reaches 0.7 dB and the phase error is approximately 5°, the six-port QPSK receiver maintains an EVM of 8.7%. This result is comfortably below the 17.5% limit mandated by modern wireless communication standards, such as LTE and 5G. These results confirm the architectural resilience of the six-port receiver, validating its effectiveness as a reliable solution for high-speed, short-range data transmission in future ultra-wideband telecommunication infrastructures. Full article
(This article belongs to the Special Issue Advances in 6G Wireless Communication Technologies)
►▼ Show Figures

Figure 1

22 pages, 477 KB  
Article
Distributed Disco Intelligent Reflecting Surfaces-Based Fully Passive Jamming for MU-MISO Systems
by Yitian Wang, Sitian Li, Huan Huang, Yanan Zhang, Luyao Sun, Yongxing Song, Jide Yuan, Tianqi Yu and Yi Cai
Electronics 2026, 15(10), 2033; https://doi.org/10.3390/electronics15102033 - 10 May 2026
Viewed by 427
Abstract
Maliciously deployed disco intelligent reflecting surfaces (DIRSs) introduce active channel aging (ACA) to achieve fully passive jamming without requiring channel state information or jamming power. To enhance this capability, we propose a distributed DIRS framework for downlink multi-user multiple-input single-output (MU-MISO) systems. By [...] Read more.
Maliciously deployed disco intelligent reflecting surfaces (DIRSs) introduce active channel aging (ACA) to achieve fully passive jamming without requiring channel state information or jamming power. To enhance this capability, we propose a distributed DIRS framework for downlink multi-user multiple-input single-output (MU-MISO) systems. By distributing multiple panels, this framework increases independent reflection paths and introduces inter-panel cascaded reflections, severely exacerbating precoder mismatch. We develop a comprehensive near- and far-field cascaded channel model, deriving closed-form expressions for the interference variance and a sum-rate lower bound in the large-antenna regime. Both pilot training (PT) phase-on and phase-off scenarios are investigated to evaluate the jamming impact under different operational states. Analytical and simulation results reveal that DIRS-induced interference scales with transmit power, imposing a strict rate ceiling. Specifically, at 10 dBm transmit power per LU, the proposed framework not only reduces the achievable sum-rate by over 57% relative to the interference-free scenario, but also improves the jamming impact by approximately 36% compared to the conventional single-panel DIRS, demonstrating superior and robust fully passive jamming capability. Full article
(This article belongs to the Special Issue Massive MIMO and Other Advanced Antenna Techniques for 5G/6G Wireless Systems)
►▼ Show Figures

Figure 1

20 pages, 1186 KB  
Article
Radio Frequency Resonate and Fire (RF-RAF) Neurons Supporting Device Classification
by David L. Weathers, Michael A. Temple and Brett J. Borghetti
Electronics 2026, 15(10), 2023; https://doi.org/10.3390/electronics15102023 - 9 May 2026
Cited by 1 | Viewed by 516
Abstract
Radio Frequency Fingerprinting (RFF) enables passive physical-layer device authentication by exploiting unintentional hardware variations in wireless transmitters. Neuromorphic implementations are attractive, given their potential for low-latency, energy-efficient inference capability under Size, Weight, and Power (SWaP) constraints at the edge. A new RFF capability [...] Read more.
Radio Frequency Fingerprinting (RFF) enables passive physical-layer device authentication by exploiting unintentional hardware variations in wireless transmitters. Neuromorphic implementations are attractive, given their potential for low-latency, energy-efficient inference capability under Size, Weight, and Power (SWaP) constraints at the edge. A new RFF capability is demonstrated here using recently introduced Radio Frequency Resonate-and-Fire (RF-RAF) neurons and eight WirelessHART devices. Performance is evaluated for RF-RAF-generated fingerprints against the established Gabor Transform (GTX) baseline using three classifier architectures: Random Forest (RndF), Convolutional Neural Network (CNN), and a Time-Incremented Spiking Neural Network (TI-SNN). The results show that RF-RAF fingerprints achieve an average classification accuracy of 96.7% across all three classifier types and consistently outperform GTX fingerprints at all evaluated fingerprint sizes. This performance persists under time-span-matched conditions, and the RF-RAF versus GTX benefit is not solely attributable to input data utilization. The TI-SNN surpasses 94% classification accuracy using M = 4 time step RF-RAF fingerprints with approximately 100 spikes per inference—a 4× larger GTX fingerprint requires approximately 1000 spikes to achieve the same classification accuracy. RF-RAF fingerprints offer two additional benefits: they are natively non-negative, which supports efficient neuromorphic hardware implementation, and they provide greater flexibility in fingerprint size selection. It is concluded that RF-RAF neurons provide an efficient neuromorphic-native encoding pathway for device RFF discrimination and offer improved accuracy–efficiency tradeoffs in training and inference for various classifier architectures. Full article
(This article belongs to the Special Issue Advances in 5G and Beyond Mobile Communication)
►▼ Show Figures

Figure 1

37 pages, 2297 KB  
Review
Space Photovoltaics: Materials, Device Concepts and Operational Challenges
by Anna Drabczyk, Paweł Uss, Katarzyna Bucka, Wojciech Bulowski, Patryk Kasza, Grzegorz Putynkowski and Robert P. Socha
Electronics 2026, 15(10), 1978; https://doi.org/10.3390/electronics15101978 - 7 May 2026
Cited by 2 | Viewed by 1937
Abstract
Space photovoltaics remains the primary power source for satellites and spacecraft, where high efficiency, radiation resistance, and low mass are essential requirements. While conventional III–V multijunction solar cells currently represent the technological benchmark, recent advances in materials science and device architectures have significantly [...] Read more.
Space photovoltaics remains the primary power source for satellites and spacecraft, where high efficiency, radiation resistance, and low mass are essential requirements. While conventional III–V multijunction solar cells currently represent the technological benchmark, recent advances in materials science and device architectures have significantly expanded the design space of space photovoltaic systems. This review provides a comprehensive overview of the fundamental physical principles, material platforms, and device concepts relevant to photovoltaic operation under space conditions, with particular emphasis on the AM0 spectrum, radiation effects, and thermal cycling. Special attention is devoted to advanced architectures, including inverted metamorphic multijunction solar cells, concentrator photovoltaic systems, and emerging tandem concepts such as perovskite/silicon and all-perovskite devices. The review highlights the growing importance of system-level metrics, particularly specific power and integration flexibility, which increasingly complement efficiency as key performance indicators. Although emerging technologies offer unprecedented opportunities for lightweight and high-efficiency photovoltaic systems, challenges related to long-term stability, defect control, and scalability remain critical for their practical implementation. Overall, the future of space photovoltaics lies in the development of application-specific solutions that balance efficiency, durability, mass, and cost, enabling next-generation space missions and energy systems. Full article
(This article belongs to the Special Issue Recent Advances in Emerging Semiconductor Devices)
►▼ Show Figures

Figure 1

22 pages, 597 KB  
Article
Scaling Computer Vision: A Comparative Analysis of Cloud Infrastructures for AI-Based Image Processing and Classification Applications
by Haojie Zheng, Carlos Reaño, Alberto Castillo, Juan F. Ariño-Sales, Álvaro Igual and Carles Igual
Electronics 2026, 15(9), 1953; https://doi.org/10.3390/electronics15091953 - 5 May 2026
Viewed by 608
Abstract
Artificial intelligence-driven computer vision has undergone rapid expansion in recent years, largely propelled by progress in deep learning techniques and the availability of extensive annotated datasets. Nevertheless, the large-scale adoption of such systems remains challenging for many organizations due to financial constraints and [...] Read more.
Artificial intelligence-driven computer vision has undergone rapid expansion in recent years, largely propelled by progress in deep learning techniques and the availability of extensive annotated datasets. Nevertheless, the large-scale adoption of such systems remains challenging for many organizations due to financial constraints and technological complexity. In this context, cloud computing has become an appealing alternative, as it offers elastic, on-demand resources under a pay-as-you-go model. Despite these advantages, the use of cloud platforms also introduces specific challenges for computer vision applications. One of the key open issues concerns the assessment of whether it is better to use classical Infrastructure (IaaS) or Containers (CaaS) to build applications. In this paper, we evaluated and compared these two models by using a real-world use case: an AI-based image processing and classification application. The best-performing model achieved speed-ups of up to 2.12× and reduced resource consumption and costs by up to 22% compared with the other evaluated alternatives. Full article
(This article belongs to the Special Issue AI-Based Image Processing Detection and Classification Analysis for Multidisciplinary Approaches)
►▼ Show Figures

Figure 1

16 pages, 919 KB  
Article
A Comparative Performance Study of Host-Based Intrusion Detection Using TextRank-Based System Call Preprocessing and Deep Learning Models
by Hyunwook You, Chulgyun Park, Dongkyoo Shin and Dongil Shin
Electronics 2026, 15(9), 1856; https://doi.org/10.3390/electronics15091856 - 27 Apr 2026
Viewed by 620
Abstract
Host-based intrusion detection systems (HIDSs) can address the limitations of network-based detection by analyzing system calls and other low-level events. Many existing benchmark datasets remain inadequate for evaluating modern attacks because they were built in outdated environments and cover only a limited set [...] Read more.
Host-based intrusion detection systems (HIDSs) can address the limitations of network-based detection by analyzing system calls and other low-level events. Many existing benchmark datasets remain inadequate for evaluating modern attacks because they were built in outdated environments and cover only a limited set of attack behaviors. To address this gap, this study builds a TextRank-based preprocessing pipeline on the LID-DS 2021 dataset and compares five end-to-end pipelines: Random Forest (RF), Long Short-Term Memory (LSTM), Convolutional Neural Network(CNN) + LSTM, LSTM, Bidirectional LSTM (BiLSTM), and CNN + Bidirectional Gated Recurrent Unit (BiGRU). Of the 15 scenarios in the dataset, six multi-stage attacks were excluded, and three representative scenarios were selected based on attack-category coverage and suitability for single-chunk host-level detection. Within these three selected scenarios and same-scenario file-level splits, the deep learning pipelines achieved F1-scores of 0.90–0.94, whereas RF ranged from 0.55 to 0.63. Among the evaluated pipelines, CNN + BiGRU produced the strongest overall results. These findings indicate that, under this constrained evaluation setting, sequential deep learning pipelines can be effective for scenario-specific system-call-based HIDS; however, broader generalization to unseen attacks or to the full LID-DS 2021 scenario set remains unverified. 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

14 pages, 1630 KB  
Article
Broadband Stepped-Impedance Wilkinson Power Divider with Improved Performance
by Stelios Tsitsos, Maria Prousali and Hristos T. Anastassiu
Electronics 2026, 15(9), 1839; https://doi.org/10.3390/electronics15091839 - 26 Apr 2026
Viewed by 688
Abstract
Herein, we present the analysis, design, optimization, and fabrication of a broadband, stepped-impedance Wilkinson power divider. The proposed structure employs stepped-impedance transmission lines and open-circuited stubs, achieving a simple and compact implementation while maintaining a wideband frequency response. Initially, transmission-line-based circuit analysis was [...] Read more.
Herein, we present the analysis, design, optimization, and fabrication of a broadband, stepped-impedance Wilkinson power divider. The proposed structure employs stepped-impedance transmission lines and open-circuited stubs, achieving a simple and compact implementation while maintaining a wideband frequency response. Initially, transmission-line-based circuit analysis was performed to extract the design equations, followed by simulation and optimization to enhance impedance matching and output-port isolation over a broad bandwidth. Finally, the proposed divider was fabricated using microstrip-line technology, and experimental measurements were conducted using the Agilent E5071C vector network analyzer. The simulation and measurement results showed efficient wideband operation over the 1–4 GHz frequency range. Specifically, the measured return loss at the input port was <−10 dB; the corresponding return loss at the output ports was <−15 dB. The measured insertion loss was −3.73 ± 0.42 dB. The isolation between the output ports was <−10 dB, reaching approximately −30 dB at 2.1 GHz and −25 dB at the center operating frequency (f0 = 2.5 GHz). The amplitude and phase imbalances were 0 ± 0.2 dB and 0o ± 0.8o, respectively. Furthermore, the overall size of the proposed wideband Wilkinson power divider was 0.35λg × 0.21λg. Compared to previous designs, the divider proposed in this study exhibits an improved and more symmetric frequency response, as well as a substantially reduced size, making it suitable for several modern wireless technologies such as Wi-Fi, Bluetooth, GPS, DCS, WCDMA, and sub-6 GHz 5G communication systems. Full article
(This article belongs to the Special Issue High-Frequency Electromagnetic Technologies for 5G/6G Communications: Antenna Design and Signal Optimization)
►▼ Show Figures

Figure 1

47 pages, 5474 KB  
Review
Bias in Large Language Models: Origin, Evaluation, and Mitigation
by Yufei Guo, Muzhe Guo, Juntao Su, Zhou Yang, Mengqiu Zhu, Hongfei Li, Mengyang Qiu and Shuo Shuo Liu
Electronics 2026, 15(9), 1824; https://doi.org/10.3390/electronics15091824 - 24 Apr 2026
Cited by 6 | Viewed by 2027
Abstract
Large language models (LLMs) have revolutionized natural language processing, but their susceptibility to biases poses significant challenges. This comprehensive review examines the landscape of bias in LLMs, from its origins to current mitigation strategies. We categorize biases as intrinsic and extrinsic, analyzing their [...] Read more.
Large language models (LLMs) have revolutionized natural language processing, but their susceptibility to biases poses significant challenges. This comprehensive review examines the landscape of bias in LLMs, from its origins to current mitigation strategies. We categorize biases as intrinsic and extrinsic, analyzing their manifestations in various natural language processing (NLP) tasks. The review critically assesses a range of bias evaluation methods, including data-level, model-level, and output-level approaches, providing researchers with a robust toolkit for bias detection. We further explore mitigation strategies, categorizing them into pre-model, intra-model, and post-model techniques, highlighting their effectiveness and limitations. Ethical and legal implications of biased LLMs are discussed, emphasizing potential harms in real-world applications such as healthcare and criminal justice. By synthesizing current knowledge on bias in LLMs, this review contributes to the ongoing effort to develop fair and responsible artificial intelligence (AI) systems. Our work serves as a comprehensive resource for researchers and practitioners working towards understanding, evaluating, and mitigating bias in LLMs, fostering the development of more equitable AI technologies. Full article
(This article belongs to the Special Issue Data Mining in Natural Language Processing: Latest Advances and Prospects)
►▼ Show Figures

Figure 1

37 pages, 7662 KB  
Article
Joint Congestion Control Evaluation for MPTCP and MPQUIC over Multi-Link Backhauls with eMBB and mMTC-like Traffic
by Roberto Picchi and Daniele Tarchi
Electronics 2026, 15(9), 1797; https://doi.org/10.3390/electronics15091797 - 23 Apr 2026
Viewed by 501
Abstract
Multi-link terrestrial backhauls create a shared transport environment in which heterogeneous multipath protocols compete for the same forwarding resources while reacting to congestion with different control logics. In this paper, we investigate this problem in a 5G Integrated Access and Backhaul (IAB) scenario [...] Read more.
Multi-link terrestrial backhauls create a shared transport environment in which heterogeneous multipath protocols compete for the same forwarding resources while reacting to congestion with different control logics. In this paper, we investigate this problem in a 5G Integrated Access and Backhaul (IAB) scenario where an IAB node aggregates traffic from multiple User Equipments (UEs) and forwards it toward the core network over two terrestrial backhaul paths. We focus on the coexistence of Multipath TCP (MPTCP) and Multipath QUIC (MPQUIC), evaluating how cross-protocol Congestion Control (CC) pairings affect performance. Specifically, all feasible BBR, CUBIC, and Reno cross-pairings are assessed under symmetric and asymmetric dual-backhaul conditions, considering Enhanced Mobile Broadband (eMBB) and dense low-rate traffic regimes representative of mMTC-like operation. The analysis considers throughput, Jain’s fairness index, jitter, and packet loss to identify the trade-offs of each CC pairing. Results show that CC selection is a first-order design factor in MPTCP/MPQUIC coexistence over shared backhauls. No single pairing is uniformly optimal across all metrics: some configurations provide more balanced throughput sharing, others improve fairness, while the most favorable solutions for jitter do not necessarily maximize transport efficiency. These findings identify CC pairing as a tuning dimension for multi-link backhaul systems based on heterogeneous multipath transports. Full article
(This article belongs to the Section Computer Science & Engineering)
►▼ Show Figures

Figure 1

Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying article 1-50 on page 1 of 21.
Go to page     1 2 3 4 5 chevron_right last_page
Back to TopTop