Sensors doi: 10.3390/s26123933

Authors: Yudong Sun Zijing Zhang Yixuan Li Shaoyang Ding Hanbo Chen Zhengeng Xu Yuejing Li Xincheng Li Dafu Wang Jun Ren

Simultaneous multi-ion detection is important for interpreting leaching, corrosion, hydration, and solidification processes in cement-based materials, because these processes are controlled by coupled ion migration, binding, and precipitation–dissolution reactions. Conventional methods such as pore-solution extraction, ion chromatography, inductively coupled plasma optical emission spectroscopy, and single-ion potentiometric measurements provide useful chemical information, but they generally rely on discrete sampling or isolated ion channels and therefore have limited ability to capture time-aligned multi-ion evolution. In this study, an IoT-based in situ multi-ion detection system was developed by integrating ion-selective electrodes for Cl−, Ca2+, F−, and H+ with an ADS1115 analog-to-digital converter, an ESP32 microcontroller, and a voltage amplification module. The system achieved minimum resolvable concentrations of 10−5 M for Cl− and F− and 10−4 M for Ca2+, while maintaining pH measurement over the range of 2–12. Ten consecutive measurements at 0.01 M showed relative standard deviations below 0.12%, indicating good short-term repeatability under laboratory calibration conditions. Interference and temperature tests showed that Br− and NO3− affected the chloride channel at high concentrations, Ca2+ reduced free F− activity through Ca–F precipitation equilibrium, and the temperature drift of Cl− and F− electrodes changed direction with concentration, whereas the Ca2+ response decreased monotonically with increasing temperature. When applied to phosphogypsum–cement hardened pastes, the system captured rapid Ca2+ release, low-level F− fluctuation controlled by Ca–F interaction, non-monotonic Cl− release, and alkaline pH evolution on the same time axis. Compared with existing single-ion or offline methods, the proposed system provides synchronized in situ evidence for interpreting coupled ion leaching in cement-based solid-waste systems.:

Sensors doi: 10.3390/s26123932

Authors: Oskar Ika Adi Nugroho Wen-Nung Lie

Skeleton-based human action recognition (HAR) requires models that preserve the local kinematic structure of the human body while capturing long-range spatiotemporal dependencies under noisy or incomplete joint observations. Traditional Graph Convolutional Networks (GCNs) provide topology-aligned inductive bias but are often limited by local information aggregation from neighboring joints. In contrast, attention-based mechanisms capture global interactions, yet they may attend to spurious correlations when skeletal constraints are weakly enforced. This paper proposes Differential Hyperedge Attention-enhanced GCN (DHA-eGCN), a hybrid architecture that couples structure-aware Differential Hyperedge Attention with multi-scale temporal convolution for spatiotemporal skeleton sequence processing. DHA injects skeletal structure into attention via hop-distance relative positional encoding and hyperedge context tokens generated via joint-to-part pooling. It further employs differential attention to suppress shared noisy correlations and enhance interaction selectivity. To strengthen spatial grounding, an explicit GCN branch is added under partial- or full-depth configurations, where the first four or all ten layers are applied with graph convolutions. The model further employs an ensemble strategy that combines predictions from multiple complementary model instances. Our experiments on NTU RGB+D 60 under the X-Sub and X-View protocols, NTU RGB+D 120 under the X-Sub and X-Set protocols, and Northwestern-UCLA demonstrate that DHA-eGCN consistently outperforms or remains competitive with strong graph-based, transformer-based, and hybrid state-of-the-art methods based on the same four-stream architecture. The best configuration achieves 93.7% and 97.0% on NTU RGB+D 60 X-Sub and X-View, respectively; 90.9% and 91.9% on NTU RGB+D 120 X-Sub and X-Set, respectively; and 97.6% on Northwestern-UCLA.

Sensors doi: 10.3390/s26123931

Authors: Hailang Huang Meng Li Jiebao Zhang Yitong Li

Detecting objects in drone-captured aerial imagery is particularly formidable due to challenges such as the prevalence of numerous small targets and their dense spatial distribution. To bridge this gap, this paper introduces YOLO-UTD (YOLO-UAV Traffic Detection), a dedicated small object detector tailored for drone traffic surveillance. Built upon the YOLOv8 framework, the proposed model incorporates three principal enhancements. First, a specialized small-object detection head replaces the original large-object head to increase the sensitivity to fine-grained features. Second, we introduce a shallow-augmented feature pyramid network (SFPN) into the neck module. The SFPN enriches the semantic content of high-resolution shallow features via dense multiscale interactions and CARAFE upsampling, boosting performance on small targets. Finally, a C2fA layer is integrated into the deep backbone stages to adaptively fuse spatial details and semantic context through a dual-path architecture and a cross-attention mechanism, thereby dynamically refining features critical for small objects. Extensive experiments on the VisDrone2019 dataset validate that YOLO-UTD achieves a 3.6% higher mean average precision (mAP) than YOLOv8 while preserving a low parameter footprint, with a particularly significant gain of 5.3% in vehicle detection accuracy. These findings confirm the model’s efficacy and strong potential for application in smart city drone surveillance.

Sensors doi: 10.3390/s26123930

Authors: Dokhyl AlQahtani Emad A. Mohamed

Small unmanned aerial vehicles are difficult short-range radar targets because their millimeter-wave radar cross-sections often fall between −10 and −25 dBsm. This paper presents a modular analytical and simulation-based benchmark of a cascaded 77 GHz TDM-MIMO FMCW radar with 12 transmitters and 16 receivers, yielding a 192-element virtual ULA over a 40 m instrumented range. The framework is organized around the main counter-UAV sensing functions: range–Doppler processing first evaluates target observability and provides range–Doppler gates; Doppler-dependent TDM phase compensation is then required before virtual-array snapshots are formed for DoA estimation; and a separate long-dwell single-transmitter branch evaluates micro-Doppler separability using handcrafted features and a nearest-centroid Mahalanobis classifier. Four benchmarks are considered: detection under Swerling fluctuation models, residual TDM phase error caused by Doppler quantization, DoA estimation under an idealized far-field snapshot model, and micro-Doppler separability among UAV and bird classes. Under Swerling I, targets with a mean RCS of −10 dBsm or larger maintain detection probability above 0.9 throughout the 40 m window, whereas the −20 and −25 dBsm classes fall below that level at about 28 m and 21 m. In the far-field DoA benchmark, TLS-ESPRIT gives the lowest conditional RMSE and remains about 13–14 dB above the subarray CRLB at moderate SNR; however, these angular results are reference ceilings because the short-range operating region violates the full-aperture far-field condition and because residual TDM phase error can be severe without accurate compensation. In the micro-Doppler benchmark, birds exceed 95% per-class accuracy at 20 dB total SNR, but overall four-class accuracy saturates near 72–75% and UAV-only three-class accuracy near 63%, with most confusion between the micro-quadrotor and fixed-wing classes. This study therefore identifies architecture-specific performance margins and limitations before measured-data field validation, rather than claiming complete deployment-level performance.

Sensors doi: 10.3390/s26123929

Authors: Hao Geng Yingjian Xie Junlong Wu Zhihao Wang Hu Han Dong Yang

Full-duplex acoustic telemetry is important for real-time bidirectional measurement and control in intelligent coiled-tubing operations, but its reliability in deep horizontal wells is limited by long-range dispersion, asymmetric flow-induced noise, and severe near-end self-interference. This study proposes an orthogonal frequency-band planning and synergistic interference suppression method for full-duplex acoustic communication in coiled tubing. A dispersion model and an asymmetric attenuation model were first established for a fluid-filled coiled-tubing cylindrical-shell waveguide to characterize the physical transmission constraints. A multiphysics multi-objective cost function was then formulated by considering dispersion flatness, channel attenuation, asymmetric noise adaptability, and spectral isolation, and an improved simulated annealing algorithm was used to optimize the uplink and downlink frequency bands. In addition, a three-stage suppression architecture integrating mechanical decoupling, physical-layer frequency isolation, and CEEMDAN–wavelet denoising was developed to reduce self-interference and residual nonstationary noise. Full-scale experiments on a 457.2 m coiled-tubing surface circulation system showed that the proposed method improved the output signal-to-interference-plus-noise ratio from −15 dB to 18.5 dB and maintained a bit error rate below 1.2 × 10−4 at 400 L/min. These results indicate that the proposed approach can enhance the robustness of full-duplex acoustic telemetry under strong flow-induced noise.

Sensors doi: 10.3390/s26123928

Authors: Claudia Campolo Alessandro Confido Domenico Gioffrè Antonella Molinaro Bruno Pizzimenti Giuseppe Ruggeri Domenico Mario Zappalà

Accurate road-event detection and timely alert message dissemination are essential for the safety of connected and automated vehicles. In many scenarios, alert messages must reach not only nearby vehicles but also remote stakeholders, such as traffic management centers, cloud services, and infrastructure operators. This requirement motivates the adoption of cellular-based communication technologies in addition to short-range vehicle-to-everything (V2X) communications for data dissemination. In this work, we investigate vehicle-to-network-to-everything (V2N2X) communications for the dissemination of alert messages generated after the on-board detection of hazardous road events through machine learning (ML) algorithms. Although V2N2X connectivity is well suited for extending data dissemination beyond the local vehicular environment, its capability to guarantee prompt message delivery under strict latency constraints remains an open challenge, particularly when ML inference is integrated into the end-to-end processing pipeline. To address this issue, we develop and experimentally evaluate a proof-of-concept (PoC) platform that combines real-time road-event detection with relevant message dissemination towards both nearby and remote recipients. The proposed framework leverages 5G connectivity and publish/subscribe messaging protocols. The experimental results showcase that dissemination latency is highly influenced by both the adopted type of 5G deployment (private versus commercial networks) and the load conditions at the message broker.

Sensors doi: 10.3390/s26123926

Authors: Maria Rashidi Shayan Ghazimoghadam Vahid Mousavi Sattar Dorafshan Behruz Bozorg

Bridge decks are continuously subjected to high environmental exposure, traffic loading, and material aging, leading to progressive delamination which can negatively affect structural integrity and public safety. More specifically, subsurface delamination of concrete and corroded steel reinforcement must be repaired to keep the decks operational. Among non-destructive evaluation techniques, Ground-Penetrating Radar (GPR) and Infrared Thermography (IRT) offer complementary capabilities for detecting subsurface and near-surface defects; however, effective GPR-IRT data fusion remains challenging due to fundamental differences in sensing principles, spatial resolution and sensitivity. This study introduces a Physics-Enhanced Multi-Modal Fusion (PE-MMF) framework that integrates GPR and IRT data to improve delamination detection in reinforced concrete bridge decks. The proposed approach leverages transfer learning, cross-modal attention mechanisms, and gated fusion to enable robust learning from heterogeneous sensor inputs. Furthermore, a systematic feature selection protocol is integrated to identify physically meaningful indicators that remain consistent across different bridges, enhancing generalization capability. The framework is trained and validated using the publicly available SDNET2021 dataset, comprising co-registered GPR and IRT measurements from five in-service bridge decks with verified delamination ground truth. Results demonstrate substantial performance improvements, with average F1-score gains of up to 55% over IRT-based methods and 25% over GPR-based methods across all tested bridges. Comparative analysis against state-of-the-art methods confirmed the superior generalization capability of the proposed multi-modal approach over single-modality approaches. The findings highlight the potential of deep learning-based sensor fusion as a scalable and data-efficient decision-support tool to prioritize regions for detailed physical investigation during long-term infrastructure monitoring.

Sensors doi: 10.3390/s26123927

Authors: Shuai Liang Xiao Wang Jialong Sun Hui Liu Huilei Yang

Remote sensing object detection remains challenging because objects often appear with large scale variation, dense spatial layouts, and strong interference from complex geographical backgrounds. To address these coupled difficulties, we propose EDM-Net, an end-to-end multi-scale detector that organizes feature processing into three coordinated stages: adaptive extraction, intra-scale interaction, and cross-scale fusion. First, an efficient sparse mixture-of-experts (ES-MoE) module is embedded in the backbone to allocate scale-specific convolutional experts according to scene-level feature responses, providing a more adaptive feature basis than a single static extraction path. Second, a dynamic mixing intra-scale feature interaction (DMIFI) module is introduced into the Transformer encoder. This module combines global self-attention with dynamic spatial mixing, thereby preserving long-range context while reintroducing local two-dimensional inductive bias for dense and small objects. Third, a multi-scale synergistic attention fusion (MSAF) module aligns adjacent feature levels through parallel local and global attention branches and structural re-parameterization, reducing semantic dilution during feature aggregation. Comprehensive experiments on three large-scale remote sensing benchmark datasets, DIOR, NWPU VHR-10, and RSOD, demonstrate that EDM-Net consistently improves over the re-trained RT-DETR-R18 baseline under the same experimental protocol, attaining mAP50 scores of 83.7%, 95.6%, and 95.8% respectively. Additional ablation and scale-specific analyses indicate that the three modules contribute complementary gains, especially for small and densely distributed objects. These results suggest that coordinated extraction, interaction, and fusion can improve remote sensing object detection under complex scale and background conditions.

Sensors doi: 10.3390/s26123925

Authors: Amani Atiani Mohammed El-Absi Thomas Kaiser

This paper investigates the feasibility of 3D self-localization and tracking using chipless radio frequency identification (RFID) tags operating in the terahertz (THz) frequency band. The primary objective is to achieve sub-millimeter (sub-mm) localization and tracking accuracy while minimizing reliance on external infrastructure. To this end, a hybrid localization framework is proposed that jointly exploits round-trip time-of-flight (RToF) and angle-of-arrival (AoA) measurements to enhance localization performance. Although near-field propagation effects are inherently significant in the considered THz operating regime, a simplified far-field approximation is adopted to facilitate tractable system modeling and analytical development. The proposed framework is further extended to dynamic scenarios through an extended Kalman filter (EKF)-based tracking algorithm, which incorporates temporal state evolution to improve estimation robustness under noisy measurements. Furthermore, the Cramér–Rao lower bound (CRLB) for the hybrid RToF-AoA system is derived to establish the fundamental limits of localization accuracy under varying system configurations and measurement conditions. Simulation results demonstrate that the proposed approach is capable of achieving sub-mm localization and tracking accuracy with a highly constrained anchor infrastructure, including operation with a single anchor in the considered scenario. These findings highlight the potential of THz chipless RFID technology as a promising enabling solution for next-generation high-accuracy localization and tracking applications.

Sensors doi: 10.3390/s26123923

Authors: Lukas Peterson Andrei Zagrai ThankGod Nwokocha T. David Burleigh

Ultrasonic thickness resonance can be effectively used to detect and quantify the level of corrosion in steel nuclear storage containers as well as other corrosion-prone thin-walled structures, such as pipes and storage tanks. Electro-Magnetic Acoustic Transducers (EMATs) have several advantages over more traditional piezoelectric-based transducers; namely, they can be used in a non-contact fashion on robotic platforms, allowing for measurements regardless of surface conditions or temperature. The major challenge of EMAT application is the power required to counteract the low actuation efficiency, which is achieved with a high-power ultrasonic pulse generator and a transformer circuit. Resonance techniques, in which most of the energy is concentrated near structural resonance frequencies, are preferable to improve efficiency of electro-magnetic acoustic measurements. This methodology was applied to 316L stainless steel thin plates subjected to uniform corrosion as well as pitting corrosion imitating different damage scenarios in a nuclear waste container. The resonant peak frequency shift was found to be proportional to the severity of corrosion for minimally corroded samples. However, the complete disappearance of the resonance peak was observed in the samples with severe corrosion damage. The EMAT liftoff distance was studied to quantify its effect on the amplitude, spread, and frequency of resonant peaks. Recommendations for use of EMATs for assessing corrosion damage are presented. The study demonstrates the success of frequency-based detection of corrosion damage in 316L stainless steel used in fabrication of nuclear waste storage containers.

Sensors doi: 10.3390/s26123924

Authors: Luigi Bibbò Filippo Laganà Salvatore A. Pullano Giovanni Angiulli

Integrating neural and muscular signals into wearable robotics enables adaptive assistance during real-world tasks. This study proposes a multimodal neural interface for passive exoskeletons that combines electroencephalography (EEG) and electromyography (EMG) signals to classify motor gestures and estimate real-time cognitive and muscular effort, supported by finite-element-based biomechanical modeling. The system was implemented on the Ottobock Shoulder X passive exoskeleton© and validated using synchronous EEG–EMG acquisition via the LiveAmp platform©, a commercially available platform that was not developed specifically for this study. A hybrid CNN–LSTM architecture with deep fusion was employed to enhance robustness and responsiveness under realistic operating conditions. This study proposes a multimodal neural interface for the software-level adaptive assistance of passive upper-limb exoskeletons. While the physical device maintains a static mechanical profile, the proposed digital framework achieves adaptation by interpreting the user’s physiological and motor states. Ten healthy participants performed three functional tasks (screwing, moving the box, and lifting the box) under five assistive conditions. Finite element modeling (FEM) was used to characterize the torque–angle relationship of the passive exoskeleton and to support the interpretation of experimentally observed assistive torque profiles. The FEM model, used as an offline biomechanical analysis tool to aid in the interpretation of experimental results, has not been integrated into the real-time control loop. Results showed an average classification accuracy of 90%, an F1-score of 0.85, and inference latency below 180 ms, confirming real-time applicability. Cognitive indices such as the Cognitive Load Index (CLI) and Frontal Asymmetry Index (FAI) enabled adaptive modulation of assistance strategies without requiring active actuation, thereby preserving the device’s intrinsic passive nature. Comparative torque analysis highlighted the ergonomic benefits of passive systems in mid-range postures, while Finite Element Method (FEM) supported analysis clarified their limitations under highly dynamic loads compared to active solutions. These findings advance multimodal brain–machine interfaces for wearable robotics by integrating physiological sensing, deep learning, and biomechanical modeling, offering a safe, energy-efficient, and adaptive approach with potential rehabilitation, occupational ergonomics, and human–robot applications.

Sensors doi: 10.3390/s26123922

Authors: Aydın Boyar Ersan Kabalcı

There is a growing trend towards vehicles powered by alternative energy sources due to the environmental pollution caused by fossil fuel vehicles. Electric vehicles (EVs) are thought to make a significant contribution to reducing environmental pollution. This study presents a performance comparison of field-oriented control (FOC) and finite control set-based model predictive current control (FCS-MPCC) methods for controlling PMSM motors, which are commonly preferred for EV applications. A multilevel ANPC inverter topology, which has a higher-quality power flow than classical two-level inverters, was preferred to power the PMSM. While the classical FOC method has a fixed switching frequency by including cascaded PI controllers and a pulse width modulation (PWM) modulator, the FCS-MPCC method determines a variable frequency-switching signal that minimizes the cost function by predicting the future current behavior of the PMSM using the mathematical model of the system. The performance comparison of FOC and FCS-MPCC methods was carried out by conducting real-time experimental studies. Both control algorithms were analyzed under variable speed and load conditions using the same motor and drive structure. Performance analysis of FOC and FCS-MPCC control algorithms was carried out in terms of speed tracking, torque, current, and harmonics. According to the results obtained, the total harmonic distortion (THD) value of the stator current was 7.03% in the FOC method, while it was 22.19% in the FCS-MPCC method. Furthermore, a comparative analysis was conducted on the dynamic performance of the two methods in different scenarios using the mean absolute error (MAE), root mean square error (RMSE), integral absolute error (IAE), integrated time absolute error (ITAE), and integral squared error (ISE) criteria. The FCS-MPCC method was observed to be superior in different speed scenarios according to these criteria. In terms of processor load, it was calculated as 17.09% in the FOC method and 63.75% in the FCS-MPCC method. This study is important for determining the control strategy of PMSMs used in EV drives.

Sensors doi: 10.3390/s26123921

Authors: Łukasz Januszkiewicz Remigiusz Danych Maciej Łaski Kornelia Bendzel

This paper presents the design of an inverted-F antenna intended for integration into a smartwatch operating in the 2.4 GHz band. The antenna design addresses spatial constraints imposed by the device’s miniaturized form factor and the proximity of electronic components, including the printed circuit board, display, and battery. The influence of the user’s body on the antenna’s performance characteristics was considered during the design phase through numerical simulations employing the Finite-Difference Time-Domain (FDTD) method with a heterogeneous human body model. Simulation results and measurements of a fabricated prototype antenna are presented, demonstrating satisfactory performance in terms of impedance matching with VSWR below 1.5 in the whole band and gain of −1 dBi.

Sensors doi: 10.3390/s26123919

Authors: Žan Regoršek Andrej Žemva

In the field of stereo depth sensing, modern research predominantly prioritizes accuracy, yet inference speed remains a critical bottleneck for practical, real-time applications on resource-constrained platforms. Existing acceleration approaches often rely on lighter network architectures or runtime-specific optimizations, which may require architectural redesign, platform-specific tuning, or accuracy trade-offs. However, a common inefficiency remains in many stereo pipelines: feature extraction is typically performed using two separate forward passes, one for the left image and one for the right, even though both passes use the same network weights. We address this redundancy by concatenating the left and right images into a single combined tensor, enabling feature extraction in one batched pass while preserving the original network architecture. By reducing feature extraction time by up to 48.4%, our results demonstrate that this method accelerates the overall inference rate by 10% to 39% on average on Nvidia V100 and up to 28.4% on edge device, depending on the model architecture. This speedup is achieved at the expense of only a moderate increase in runtime memory consumption, while retaining the original accuracy. Because the method does not alter the core stereo network, it can be applied as a plug-and-play enhancement to both existing and newly developed stereo matching models.

Sensors doi: 10.3390/s26123920

Authors: Ryan Aalund Vincent P. Paglioni

IoT devices operate as integrated systems spanning hardware, firmware/software layers, and communication layers. In operational settings, many faults and performance degradations are emergent: they arise from cross-layer interactions, workload changes, and telemetry artifacts, rather than a single physics-of-failure mechanism. These realities make traditional supervised fault classification difficult because labeled fault data are rarely available during deployment, and the fault surface is unknown and a priori. This paper presents a practitioner-oriented, label-free fault detection and diagnosis (FDD) pattern based on Dynamic Time Warping (DTW) for rapid implementation in production IoT telemetry. The method represents a device as a sequence of overlapping episodes and organizes telemetry into interpretable layers (hardware sensors, communication health proxies, and software/firmware-derived KPIs). A reference library of regular episodes is built from an assumed-healthy training window; new episodes are scored using constrained DTW distances against this library, while retaining per-layer and per-channel contributions for attribution. We show that production performance depends strongly on operational parameterization, including episode length, DTW constraints, robust threshold learning, and temporal validation. Within a verified-healthy evaluation window, the tuned configuration achieves an AUROC of 0.97 for the temporally structured faults DTW is suited to (bias, drift, and interaction faults, with spikes detected at an AUROC of 0.93), detecting 100% of injected faults, with a mean delay under 25 min. We further show that constant-value (stuck-at) and missing-data (dropout) faults fall outside DTW’s shape-matching scope (AUROC about 0.66) and are better served by complementary variance- and missingness-based detectors, a consequence of DTW’s shape-matching scope rather than a parameter choice. This work contributes a system-level methodological framework for deploying DTW as an IoT fault-detection-and-diagnosis capability: an episode-and-layer architecture aligned with hardware, communication, and software/firmware ownership; a label-free reference library requiring only assumed-healthy data; per-layer and per-channel attribution for cross-domain triage; and a reproducible operational tuning procedure. Together, these deliver a fast-to-deploy, scalable, and accurate first-line detector for label-scarce IoT systems.

Sensors doi: 10.3390/s26123918

Authors: Radu-Laurențiu Roșca Andrei-Iulian Iancu Adrian Burlacu Cătălin Dosoftei

Autonomous mobile robots are used in optimizing warehouse logistics, yet achieving precise positioning during docking maneuvers and autonomous planning remains a technical challenge. This study presents a custom vision-based control system designed for an autonomous omnidirectional wheeled robot. The proposed methodology acquires visual feedback using a stereo camera integrated within the Robot Operating System framework. Two visual feedback control laws are formulated and rigorously evaluated: a Classic Position-Based Visual Servoing algorithm, which minimizes pose error using a quaternion-based approach, and a second solution that utilizes Dual Lie Algebra to compute the 3D visual sensor’s velocities, ensuring convergence towards the desired point-feature configuration. Experimental validation reveals that while both methods achieve docking, the dual pose-free approach enables more robust, effortless movement of the robot platform than Classic Position-Based Visual Servoing. Consequently, these findings indicate that integrating depth-based feature recovery with advanced algebraic strategies offers a stable control strategy for automated industrial scenarios.

Sensors doi: 10.3390/s26123917

Authors: Bowen Dong Xinyu Zhang Weiyan Zhu Lingmin Hou Chaoya Yan Yifan Feng Lixing Lin

Short-term traffic forecasting is a critical component of intelligent transportation systems. While deep learning architectures for this task have proliferated rapidly, the sensor-level data characteristics—zero-value prevalence, distributional heterogeneity, and cross-sensor correlation structure—that drive architecture-specific failure modes remain insufficiently understood, and their implications for evidence-based model selection in real deployments have not been systematically addressed. This study addresses that question through a sensor-network diagnostic framework applied to the METR-LA dataset (Metropolitan Los Angeles; 207 inductive loop detectors, 5-min resolution). The framework integrates systematic characterization of sensor data properties, a controlled benchmark of four representative architectures—Transformer, Spatio-Temporal Graph Convolutional Network (STGCN), Diffusion Convolutional Recurrent Neural Network (DCRNN), and Gated Temporal Convolutional Network (Gated TCN)—under a unified 12→3 prediction setting, and a novel per-sensor regression analysis that quantitatively links zero-value ratios to model-specific prediction errors across all 207 sensors. Building on these findings, this study further proposes Graph-Enhanced Transformer (GETFormer), a lightweight hybrid architecture that augments the Transformer with a single-hop Graph Convolutional Network (GCN) layer and a gated residual fusion module. The diagnostic findings and condition-dependent model-selection guidelines provide an empirically grounded foundation for principled hybrid architecture development in urban traffic sensing.

Sensors doi: 10.3390/s26123915

Authors: Lei Wang Tzu-Ming Hsia Chen-Wei Hsu Pin-Yi Liu Qian-Xun Hong

Surface-deployed wireless sensor networks (WSNs) provide a flexible platform for ocean monitoring, but ocean-current-dominant marine forcing causes persistent topology evolution, backbone distortion, and route breakage. This paper proposes an event-driven self-healing routing and topology-maintenance mechanism for drift-prone surface WSNs. The design combines dual-threshold cluster-head handover, CH-HELP backbone repair, Node-HELP member reattachment, loop-free upstream reselection, and conditional global reclustering as a low-frequency corrective layer for long-term topology degradation. Unlike fixed-round reorganization, the proposed framework prioritizes local repair and triggers global refresh only when backbone quality persistently deteriorates. Simulations driven by Taiwan Strait current-dominant flow–wind data show that the full Proposed-Hybrid method reduces the CH-disconnection rate from 8.15% in DARCR to 5.15%, whereas the local-only configuration without conditional global reclustering yields 9.13%. Conditional global reclustering further suppresses late-stage topology degradation, reducing the final-third mean CH-disconnection rate from 16.32% to 8.51% and the late-stage 95th-percentile peak from 34.43% to 17.21%. DARCR remains competitive in some late-stage metrics because of its fixed-period global reorganization.

Sensors doi: 10.3390/s26123916

Authors: Yuan Zheng Haisong Chen Huan Wan Yongxue Wang

This paper investigates backscatter-aided relaying for interactive dual-HAP wireless-powered sensor networks (WPSNs), in which two cooperative sensor groups transmit sensed data to opposite hybrid access points (HAPs) using harvested radio-frequency energy. Each group consists of multiple source sensor nodes (SNs) and one relay SN selected according to its proximity to the target HAP. To reduce local cooperation overhead, source SNs reuse the wireless power transfer (WPT) signal as a controllable carrier and convey their information to the relay SN through passive backscatter communication. The collected information is then delivered to the target HAPs through direct source transmission and relay forwarding. A source common-throughput maximization problem is formulated by jointly optimizing time allocation, transmit energy allocation, and dual-HAP energy beamforming, subject to energy-causality and relay minimum-rate constraints. To address the resulting non-convexity, an alternating optimization algorithm is developed, where the time-and-energy allocation subproblem is transformed into a convex form and the energy beamforming matrices are updated through energy-feasibility margin maximization. Numerical results show that the proposed scheme outperforms active cooperation without backscatter and direct transmission, demonstrating the effectiveness of integrating passive local information collection, relay-assisted uplink transmission, and optimized dual-HAP WPT.

Sensors doi: 10.3390/s26123914

Authors: Yunsung Kim Gyeongdeok An Kihyun Kim Jaecheol Ha

Industrial control systems (ICSs) are cyber–physical environments in which physical process data and network communication data are generated simultaneously. Existing studies have mainly focused on either sensor-based or network-based anomaly detection, making it difficult to capture diverse attack indicators and motivating the use of multimodal methods that can leverage complementary information from both modalities. In this paper, we propose an unsupervised multimodal anomaly detection framework for ICSs that jointly uses sensor and network modalities. For each modality, autoencoder-based single-modality models are trained in an unsupervised manner, and their anomaly scores and latent feature vectors are extracted. These outputs are temporally aligned to construct a time-aligned multimodal table, which is then used to implement and compare two fusion strategies: anomaly score fusion and latent feature fusion. In latent feature fusion, aligned modality-specific latent features are combined with canonical correlation analysis (CCA)-derived cross-modal correlation features. The experimental results showed that latent feature fusion achieved stable performance across multiple sensor–network encoder combinations. In particular, the gated recurrent unit–convolutional neural network (GRU–CNN) combination achieved the best F1-score of 0.9166 and ROC-AUC of 0.9795. In addition, the complementarity analysis showed that latent feature fusion recovered some missed detections by integrating complementary sensor and network evidence. These results demonstrate that latent feature fusion is an effective multimodal strategy for ICS anomaly detection.

Sensors doi: 10.3390/s26123913

Authors: Qingbo Du Yapeng Yang Xiaoyu Zhao Qi Lv Yuyao Tang Jiapeng He Yier Liu Guoqiang Li

The NaI(TI) detector is highly favored in gamma radiation detection and widely applied in fields such as environmental radiation monitoring, nuclear medicine, and laboratory gamma-ray spectroscopy. Its detection performance determines the results of quantitative gamma-ray detection, making it a crucial indicator in detector design and development. This study employs the Monte Carlo method and utilizes TopMC 1.0 software to establish a NaI(TI) detector model. First, the effects of crystal size, ray energy, cladding thickness, and distance on the detector’s detection efficiency were investigated. Subsequently, the energy resolution and peak-to-total ratio of the detector were simulated and calculated, with comparisons made to experimental values. The results indicate that all three detection efficiencies of the NaI(TI) detector are positively correlated with crystal size and exhibit an initial increase followed by a decrease with rising gamma-ray energy. Both the absolute detection efficiency and full-energy peak detection efficiency first increase and then decrease with increasing cladding thickness, while showing a negative correlation with detection distance. The intrinsic detection efficiency is almost unaffected by cladding thickness and initially rises before declining with increasing detection distance. The simulated values of energy resolution closely match experimental values, improving with higher gamma-ray energy. The deviation between simulated and experimental values for different source peak-to-total ratios remains within 6.25%, verifying the model’s reliability and the accuracy of simulation data. These findings provide valuable references and guidance for optimizing detection performance, conducting source-free efficiency calibration, and structural design of NaI(TI) detectors.

Sensors doi: 10.3390/s26123912

Authors: Chengdi Zhang Mingxuan Yu Wenzhang Dong Jiaxuan Zhan Shengdong Yu Jinyu Ma Mingyang Xie

In the resistor images used in this study, many defective regions are weak coating-like marks rather than obvious scratches or pits. Their appearance is close to the epoxy background, and some visible defects were missing from the original annotation files. If these labels are used directly, the detector treats the missed defects as background samples during training. We therefore corrected the supervision before changing the feature constraint. An early YOLO26s model was first used to nominate low-overlap boxes, and these candidates were then checked manually. Only confirmed defects were merged into the labels. After this step, a scale-gated prototype consistency term was added during training to reduce the model’s bias toward the dominant tiny-defect group. On the fixed corrected benchmark, mAP50 improved from 28.14% to 63.20%, and Recall increased from 18.42% to 62.20%. In the end-to-end deployment view, where the raw and cleaned validation sets answer different practical questions, mAP50 changed from 43.66% to 63.15%, and Recall changed from 30.01% to 62.24%. For normal-size defects, Recall increased from 26.09% to 56.52%. A prototype-only transfer study on the public MVTec AD benchmark further evaluates whether the feature constraint generalizes when the label-repair stage is not applicable to clean public annotations. Since the prototype term is removed after training, the deployed detector remains the original YOLO26s model without an additional inference branch.

Sensors doi: 10.3390/s26123911

Authors: Somprasonk Gabbualoy Pattarapong Phasukkit Supan Tungjitkusolmun

Background/Objectives: Deep learning models for digital mammography sensor data are increasingly deployed across hospitals using different X-ray detector technologies and patient populations. Whether models trained on one sensor platform and population maintain accuracy when transferred to another has not been tested for the latest generation of mammography-specific foundation models under one controlled protocol. Methods: We fine-tuned five backbone architectures (ResNet-50, DINOv2-B14, Rad-DINO, Mammo-CLIP B5, and Mammo-FM) on CBIS-DDSM (film-digitized, USA, n = 714 validation) with three seeds, ablated a density-aware focal loss across three auxiliary weights, and evaluated transfer to three external sensor cohorts: CMMD (full-field digital, China, n = 1032), DMID (mixed digital, India, n = 509), and MIAS (film-digitized, UK, n = 322). Significance used paired DeLong z-tests with Benjamini–Hochberg FDR correction; temperature scaling tested post hoc recalibration at all transfer targets. Results: Within this single-source three-seed evaluation, ResNet-50 outperformed all four foundation models on CBIS-DDSM (AUC 0.867 vs. 0.847, 0.846, 0.813, and 0.703; all gaps p_adj < 0.05). The density-aware focal loss degraded both AUC and calibration at every weight tested. At transfer, every model lost 0.165 to 0.320 AUC points relative to in-distribution performance, with sensitivity at 95% specificity collapsing from 0.31 to 0.47 in-distribution to 0.11 to 0.22 across the three external targets. A per-seed Stouffer meta-analysis confirms that Mammo-CLIP B5 and Mammo-FM significantly outperformed ResNet-50 on DMID and Mammo-CLIP on CMMD, after BH-FDR; MIAS comparisons remained directional only. In the extremely dense subgroup (BI-RADS D4), Mammo-FM reached AUC 0.870 versus ResNet-50 at 0.842, a directional observation whose 95% CIs overlap heavily at the n = 140 sample size and which we do not interpret as a statistically supported advantage. Conclusions: In this single training-source, three-seed protocol, mammography-specific pretraining did not deliver the in-distribution AUC premium reported in the originating papers, and no architecture reached a level at which transfer deployment without local validation would be defensible. We frame these as observations specific to the present protocol rather than as broader conclusions about foundation models for mammography classification. The findings argue for sensor-stratified and population-stratified external validation and for local recalibration as practical prerequisites before clinical use. Code and weights are released under MIT license.

Sensors doi: 10.3390/s26123910

Authors: Vasilis Vasileiadis Anastasios Theodoropoulos George Lepouras

The integration of olfactory feedback into Virtual Reality (VR) applications remains significantly underexplored compared with other sensory modalities, particularly within room-scale Cave Automatic Virtual Environments (CAVEs), where related research is even more limited. To address this gap, this paper presents Scentree, a custom olfactory system capable of delivering scents in real time based on user interactions, along with Smelling Ancient Greece, an olfactory-enhanced VR experience developed for integration within our CAVE system. Central to the proposed approach is the concept of the Diegetic Olfactory Feedback Loop, which reframes olfaction from a passive ambient effect into an active, interaction-driven feedback mechanism embedded within the narrative context of the virtual environment. To evaluate the system, we conducted a technical performance assessment and an exploratory user study (N=51) examining participant perceptions of immersion, presence, perceived realism, usability, and overall user experience. The findings support the feasibility of interaction-driven olfactory feedback as a multisensory design approach for CAVE environments and provide a foundation for future controlled investigations of olfactory feedback in immersive VR.

Sensors doi: 10.3390/s26123909

Authors: Xiaochen Li Hongtian Zhao

Object detection in low-light scenes is difficult because weak illumination reduces local contrast, amplifies sensor noise, and makes small or occluded objects hard to localize. Existing enhancement-before-detection pipelines can improve visual brightness, but they are not always optimized for detection features, while transformer-style global reasoning is often too costly for lightweight detectors. To address this gap, we propose GE-Detection, a detector-side framework that integrates Global Sub-Sampled Attention (GSA), Efficient Multi-scale Attention (EMA), and dropout regularization into YOLO- and PicoDet-style architectures. GSA introduces lower-cost global context modeling through spatially reduced key-value tokens, EMA refines multi-scale fused features without aggressive channel compression, and dropout improves training-time regularization with no inference-time parameter overhead. Experiments on COCO, ExDark, BDD100K-Night, and NightOwls show that the method is most effective in low-light detection: on ExDark with YOLO11n, mAP50-95 improves from 34.39% to 36.74%, mAP50 from 56.24% to 59.27%, and Box (P) from 67.63% to 71.36%. The full YOLO11n variant uses 2.91M parameters and maintains 134.7 FPS on an RTX 2080 Ti under the tested setting. Cross-dataset and corruption experiments further indicate that the proposed modules improve localization under several nighttime domain shifts while retaining known limitations under severe noise and adverse weather. These results indicate that combining efficient global attention, multi-scale feature recalibration, and targeted regularization can improve low-light localization while keeping the detector practical for deployment.

Sensors doi: 10.3390/s26123908

Authors: Jiajun Chen Shaochen Jiang Yongming Li Sulaiman Tuersunayi Yong Liu

Unmanned Aerial Vehicle (UAV) image object detection holds significant application value in the low-altitude economy, traffic monitoring, intelligent agriculture, and disaster rescue. However, due to the top-down perspective, UAV images typically suffer from challenges such as small target scales, dense object distribution, severe occlusions, and complex backgrounds. These issues often limit the recall and localization accuracy of general-purpose detectors when they are directly applied to UAV small-object detection scenarios. To address these aforementioned challenges, this paper proposes an Adaptive Dynamic Aggregation YOLO network, termed ADA-YOLO. The novelty of ADA-YOLO lies in its highly efficient combinatorial design specifically tailored for UAV small object detection, while retaining the efficient backbone of YOLOv8, we systematically reconstruct the neck and detection head to improve accuracy. Specifically, a high-resolution P2 detection branch is incorporated to construct a P2–P5 multi-scale prediction structure. Furthermore, the lightweight DySample dynamic upsampling module is adopted to replace traditional upsampling methods, and an Adaptive Spatial Feature Fusion (ASFF) mechanism is introduced to alleviate semantic conflicts and noise interference during multi-scale feature fusion. This synergistic combination explicitly addresses multi-scale representation challenges and enhances small-object detection performance in complex scenes. Comparative experiments with the baseline YOLOv8n on the VisDrone2019 dataset demonstrate that ADA-YOLO achieves an improvement of 11.3% in mAP@0.5 and 8.2% in mAP@0.5:0.95. The improved model achieves these performance gains with a modest parameter increase and acceptable computational complexity. Finally, ablation experiments further validate the effectiveness of each individual module and their synergistic gains.

Sensors doi: 10.3390/s26123907

Authors: Longjie Zheng Junlin Zhou Haijun Peng Bai Li Xinwei Wang

With the increasing complexity of unmanned aerial vehicle (UAV) missions in complex obstacle environments, cooperative hunting of maneuvering ground targets by UAV swarms has become an important problem for multi-agent autonomous decision-making. This paper focuses on a simulated three-UAV hunting scenario in a two-dimensional obstructed environment, where UAVs must search for, approach, encircle, and continuously track a target while avoiding static obstacles under local observation. To address the problem of multi-UAV cooperative hunting of dynamic targets in complex obstacle environments, this paper proposes a curriculum learning (CL)-based Multi-Agent Proximal Policy Optimization algorithm, termed CL-MAPPO. Specifically, a three-stage progressive training curriculum is designed to overcome the challenges of low exploration efficiency, slow environmental adaptation, and difficult convergence of cooperative hunting policies faced by multi-agent deep reinforcement learning in hunting tasks, thereby gradually enhancing the cooperative hunting capability of UAVs in complex environments. Curriculum I employs fixed obstacles and a stationary target position to train the UAVs’ basic obstacle avoidance and target search abilities. Curriculum II introduces randomly generated obstacles and target positions to improve the UAVs’ adaptability to varying environments. Curriculum III further incorporates a dynamic target, prompting the UAVs to learn effective hunting strategies against maneuvering targets. The simulation experiment includes ablation experiments against MAPPO without curriculum learning and comparative simulations against MADDPG and MADQN, using reward convergence curves and trajectory visualizations to evaluate the training results. The results show that, under the same training episodes in the ablation experiment, CL-MAPPO reaches a higher and more stable reward level than vanilla MAPPO, indicating improved learning efficiency without increasing model complexity. In the comparative experiment, the CL-MAPPO algorithm achieved a higher success rate in cooperative hunting. These simulation experiments verify the effectiveness and superiority of the CL-MAPPO algorithm in multi-agent cooperative hunting tasks.

Sensors doi: 10.3390/s26123906

Authors: Xiaonan Pan Miao Yuan Jianrui Zhang Xiaojun Yu

Optical coherence tomography (OCT) is an optical imaging modality that provides high-resolution cross-sectional imaging of biological tissues noninvasively. In Fourier-domain OCT, axial resolution is governed by both the center wavelength and the spectral bandwidth of the light source; therefore, limited or discontinuous bandwidth degrades depth resolution and introduces sidelobes and artifacts in OCT images. To address these issues in OCT image reconstruction from gapped spectra, a sparse parameter estimation approach based on Sparse Iterative Covariance-based Estimation (SPICE) is proposed in this study. By utilizing a sparse parameter estimation framework to directly resolve depth-dependent components from discontinuous interferograms, SPICE enhances axial resolution while suppressing sidelobe artifacts inherent in standard interpolation. Experiments on multi-layered tape, oral epithelium, and finger skin show that SPICE visually suppresses gap-induced sidelobe artifacts and improves structural interpretability under representative gap conditions. Quantitative evaluations on multi-layer tape and biological tissues show that SPICE reduces axial FWHM by 30–45%, increases SSIM by 0.15–0.25, and achieves significantly lower computational cost than GAPES (p < 0.01).

Sensors doi: 10.3390/s26123905

Authors: Bilie Hu Peishen Gao Yan Liu Xi Xia Guoping Huo

Traditional white canes offer a limited perception range, whereas end-to-end visual models face challenges in real-time deployment on edge devices. To address these limitations, this paper proposes a lightweight real-time assistive system that integrates geometric topology reconstruction with state-adaptive warning logic. The system utilizes YOLOv9 to extract discrete semantic primitives of tactile paving. It constructs a dual-branch perception framework based on Median Absolute Deviation and the Minimum Spanning Tree algorithm to analyze the topological structure of tactile paving. For complex intersections characterized by warning indicators, a one-dimensional connectivity clustering algorithm based on longitudinal topology is proposed. It generates accurate macroscopic feasible directional prompts under field-of-view boundary constraints. Additionally, a hierarchical scheduling framework dynamically orchestrates scenario-specific finite state machines to enable continuous dynamic interaction across typical high-risk scenarios. Evaluated on a custom real-world dataset, the system achieves a 95.21% frame-level comprehensive accuracy for straight-path deviation correction and intersection directional prompting. Dynamic temporal stress tests confirm the temporal stability and logical coherence of state transitions. Furthermore, latency evaluations demonstrate the logic layer’s minimal computational overhead, proving its theoretical feasibility for real-time edge deployment. This approach provides an effective, low-latency solution for delivering directional prompts and hazard warnings to visually impaired users.

Sensors doi: 10.3390/s26123904

Authors: Basma Mostafa Hanan Haj Ahmad Yazan Rabaiah Marwa Elseddik

Safety-critical Industrial Internet of Things (IIoT) sensor networks deployed in disaster scenarios require intelligent routing mechanisms that prioritize mission-critical packets without relying on centralized coordination. Federated learning on resource-constrained edge nodes presents three primary challenges: the absence of an interpretable supervisory signal, the inability to act conservatively based on per-inference confidence, and vulnerability to partial node availability. The proposed FedCARE framework addresses these issues by employing a Mamdani Fuzzy Inference System to generate traceable criticality labels from multi-modal sensor telemetry, a dropout-aware aggregation protocol that normalizes over only reachable nodes, and a confidence-gated resolver that defers to symbolic fuzzy classification when model confidence is insufficient, otherwise applying an auditable maximization rule to prevent under-prioritization of safety-critical data. Evaluation on 50-, 100-, and 200-node Watts–Strogatz topologies under fault rates up to 50%, using the Edge-IIoTset and WUSTL-IIoT-2021 benchmarks, demonstrates 99.00% critical recall and up to 1.8× higher overall-packet delivery compared to RPL-RP under severe fault conditions. Routing improvements are primarily attributed to fuzzy criticality labeling and multi-path replication. These findings indicate that fuzzy-supervised federated inference offers a practical and interpretable solution for safety-critical IIoT routing, with an observed energy overhead of 7.8% per delivered packet.

Sensors doi: 10.3390/s26123897

Authors: Marco Badii Giovanni Lasagni Monica Righini Giovanni Collodi Stefano Maddio Alessandro Cidronali

In this paper, we present an optimized driving strategy for a dual RF input envelope tracking power amplifier (ET PA) exploiting load modulation. The dual-input architecture enables dynamic load modulation (LM), allowing real-time adjustment of the load impedance to enhance performance over the signal dynamics typical of digital modulation schemes. The proposed approach considers a GaN HEMT-based LM-ET PA characterized under pulsed excitation across multiple amplitude and phase conditions of the load modulation control. Optimizing the control parameters yields a suitable shaping function that extends conventional ET supply modulation to include amplitude and phase control of the auxiliary amplifier, thereby improving the efficiency, output power, and linearity of the main amplifier. Experimental data demonstrate that the proposed dual RF input GaN-based LM-ET PA at 3.6 GHz outperforms a conventional ET PA in both efficiency and linearity when tested with high peak-to-average ratio (PAPR) signals.

Sensors doi: 10.3390/s26123900

Authors: Elena Stracca Olga Napolitano Lucia Pallottino Paolo Salaris

Achieving safe and efficient navigation in cluttered and dynamic environments remains an open challenge for mobile robots, especially when perception and actuation are uncertain. Standard navigation stacks typically handle obstacle avoidance through fixed safety margins or costmap inflation layers. While effective in simple settings, these approaches are difficult to tune in practice: conservative inflation can prevent traversal through narrow passages, whereas less conservative settings may lead to unsafe behavior. Moreover, they usually encode risk only as a function of obstacle proximity. We propose a unified probability-inspired risk-cost map that integrates perception uncertainty, actuation uncertainty, dynamic obstacle prediction, and occlusion-aware memory into a single spatial representation. The resulting risk map is used by a local path-modification module that adapts a reference global path using the proposed risk map and interfaces with a standard Model Predictive Path Integral (MPPI) controller. The proposed method is compatible with standard navigation pipelines. We validate the resulting framework in Gazebo simulations under different sensing and actuation uncertainty conditions and in environments containing unknown static and dynamic obstacles. The results show that the proposed method is more robust than conventional costmap-based baselines, resulting in fewer aborted goals in cluttered environments and substantially fewer collision events when dynamic obstacles are present.

Sensors doi: 10.3390/s26123903

Authors: Leandro José Duarte Andréia Coelho Domingos Alan Petrônio Pinheiro Lorenço Santos Vasconcelos Fabrício Augusto Matheus Moura Fernando Elias de Freitas Fadel Patrícia Naomi Sakai

Partial discharge (PD) source-type classification is essential for condition-based maintenance of high-voltage apparatus. Existing approaches based on grid discretizations of phase-resolved partial discharge (PRPD) patterns suffer from performance degradation under stochastic interference and multi-source conditions. This paper proposes a graph-based framework that integrates the morphological characterization of raw high-frequency PD waveforms with the phase-amplitude position of individual discharge events to enable multi-label classification, identifying multiple PD sources coexisting within a single test. The framework operates through three stages: a multi-task neural network extracts per-pulse embeddings and confidence scores; a construction procedure establishes selective graph connectivity based on spatial proximity and morphological similarity; and an edge-conditioned graph neural network performs classification via message passing weighted by multimodal edge attributes. Experimental evaluation on PD measurements acquired in accordance with IEC 60270 shows that the proposed framework achieves a Matthews correlation coefficient (MCC) of 0.98 and an exact match ratio of 0.97 across single-source, noisy, and multi-source conditions, substantially outperforming histogram- and set-based baselines. The framework maintains an MCC of 0.97 in multi-source scenarios, where its advantage over existing methods is most pronounced. Cross-domain evaluation on an independent dataset acquired with different laboratory equipment confirms the approach’s robustness, achieving an MCC of 0.93 without retraining. Finally, an ablation study demonstrates that the joint removal of morphological similarity filtering and confidence-based node filtering and edge gating reduces the MCC by 0.25, confirming the critical role of the waveform-informed relational structure.

Sensors doi: 10.3390/s26123902

Authors: Ehiabhili Prabhu Kannan

Although optical fiber-based surface plasmon resonance (SPR) sensors have revolutionized real-time, label-free biosensing, conventional designs suffer from limited multi-analyte detection capabilities. This study utilizes the novel Pi (π)-configured dual SPR optical fiber sensor with two opposing side-polished surfaces, enabling plasmonic excitation for simultaneous multi-analyte detection. The proposed sensor leverages asymmetric metallic thin films such as Ag, Au, Cu, and hybrid configurations (metal + TiO2) to generate two distinct resonance peaks, significantly enhancing detection versatility. Numerical simulations using the finite element method in COMSOL Multiphysics v6.3 demonstrate that the π-configuration achieves dual resonance dips at 982 nm and 1276 nm for Ag and Ag–TiO2 films, 1040 nm and 1317 nm for Au and Au–TiO2 films, and 977 nm and 1249 nm for Cu and Cu–TiO2 films, respectively, for an analyte refractive index of 1.42. A peak spectral separation >125 nm was achieved for all the sensors for a refractive index range of 1.37–1.42, ensuring that the two dips are resolvable since the change in SPR wavelength is greater than or equal to the full width at half maximum, preserving dual-analyte capability and minimizing potential crosstalk. The results indicate that the π-configured dual SPR sensor utilizing silver and silver–TiO2 sensing layers had the highest wavelength sensitivity of 12,600 nmRIU−1 and 20,000 nmRIU−1, respectively, slightly outperforming its gold and copper counterpart. The optimized metallic and hybrid nanostructured films ensure dual distinct peaks with high sensitivity, while maximizing refractive index resolution. This work presents the design of a π-configured SPR-based optical fiber sensor utilizing dielectric and multi-metallic thin films, thereby offering a breakthrough in multiplexed biosensing for applications in medical diagnostics, environmental monitoring, and chemical detection.

Sensors doi: 10.3390/s26123901

Authors: Di Xu Hongli Chen Yansen Zeng Yifan Yang Jinghan Huang Jiarui Song Yan Zhan

With the widespread adoption of intelligent terminals, edge devices, and distributed information systems in the financial domain, financial security sensing data exhibit multisource heterogeneity, dynamic temporal patterns, and high privacy sensitivity. Traditional centralized anomaly detection methods are no longer able to simultaneously satisfy the requirements of cross-institutional or cross-node collaborative modeling, client data privacy protection, and robust monitoring of transaction and system anomalies. To address this challenge, a data-local federated deep anomaly detection framework has been proposed for distributed financial security sensing systems. Initially, a local deep financial security sensing representation module is constructed to perform temporal encoding and attention-based modeling on multisource financial signals, including terminal operation status, network transaction communication, backend server operation, identity authentication, and anomaly alerts, thereby extracting representations relevant to anomalous behaviors. Subsequently, a data-local federated optimization and personalized aggregation mechanism is developed to enable cross-node knowledge sharing without transmitting raw transaction or client data, while local personalized detection heads are employed to adapt to non-independent and identically distributed (non-IID) financial institution data. Furthermore, an adversarially robust security detection and trust-aware aggregation strategy is introduced to enhance model stability under input noise, feature masking, anomaly camouflage, and potential malicious client updates. Experimental results demonstrate that the proposed method achieves an Accuracy of 92.37%, a Precision of 89.41%, a Recall of 88.26%, an F1-score of 88.83%, an AUC of 93.06%, and a PR-AUC of 89.15% in the primary financial anomaly detection task, significantly outperforming baseline methods such as Isolation Forest, Autoencoder, LSTM, Transformer, FedAvg, FedProx, SCAFFOLD, and MOON. In robustness experiments, the method attains F1-scores of 87.95%, 86.42%, 86.88%, 84.57%, 86.73%, and 83.91% under Gaussian noise, feature masking, temporal shift, adversarial perturbation, and 20% and 30% malicious client scenarios, respectively. Ablation studies further confirm the effectiveness of local representation learning, personalized federated optimization, adversarial training, and trust-aware aggregation mechanisms. Overall, the proposed approach provides an efficient intelligent anomaly detection solution for financial AI security monitoring scenarios characterized by data localization requirements, node heterogeneity, and attack perturbations.

Sensors doi: 10.3390/s26123899

Authors: Ricardo Moreno Jorge Ortigoso-Narro Daniel de la Prida Luis A. Azpicueta-Ruiz Borja Genovés Guzmán Marco Raiola

While state-of-the-art massive acoustic arrays typically rely on costly, specialized FPGA architectures or rigid proprietary hardware, there is a growing need for modular, high-density sensing in complex aeroacoustics environments. This paper presents the electronic and acoustic design of a multiplexed, modular, scalable, and low-cost massive acoustic array (MxArray) founded on an embedded Linux system. The AM3358 SoC microprocessor collects audio data through its multichannel audio peripheral, where it simultaneously receives four Time-Division Multiplexing streams of 16 microphones each. This multiplexed scheme enables the handling of 64 microphones per module, whose acquisition synchronization is set with the Precision Time Protocol and a pulse injection hardware. The combination of both BeagleBone Black and microphones based on Micro-Electro-Mechanical Systems yields a cost-effective solution with built-in Ethernet connectivity and accessible software development through an embedded Linux environment with audio libraries for hardware control. Sensors are arranged in an Underbrink Spiral pattern on a four-layer printed-circuit board. The perforated thin layout minimizes any airborne disturbance, exploiting a distribution that simultaneously achieves a low sidelobe level and a narrow main lobe when used with a beamforming algorithm. Measurement results for the developed module are presented, as well as an evaluation of a full-scale system comprising 16 modules (1024 microphones) arranged in a honeycomb pattern. The resulting instrument offers a practical and scalable solution for applications that require a large number of simultaneous microphone measurements, such as beamforming technology for aeroacoustics applications.

Sensors doi: 10.3390/s26123898

Authors: Dominik Zaborniak Paweł Kasza Marcin Pazera Marcin Witczak

Efficient task assignment for mobile robots is a crucial challenge in modern intralogistics. This paper presents an integrated cyber-physical framework combining predictive tree search on switching max-plus linear systems with a physical IoT-based dispatch interface. The scheduling problem is modelled as a discrete event system, where standard max-plus algebra captures robot synchronization, and a switching mechanism represents alternative resource assignments. To address real-world operational disturbances, the predictive model is enhanced with a fault-tolerant control (FTC) mechanism that dynamically estimates and adapts to non-stationary transport delays. The resulting decision space, which grows exponentially with the prediction horizon, is explored via a predictive tree search algorithm utilizing a quadratic cost function to penalize excessive and uneven transport times. The physical dispatch layer is realized using KIS.BOX IoT devices acting as operator-controlled stations, communicating with the central controller via a WebSocket/STOMP event stream and a lightweight REST API. Simulation results obtained in a Blender 3D environment demonstrate that the proposed FTC predictive strategy significantly reduces the variance of task completion times under fault conditions compared to a baseline First-In-First-Out approach. Furthermore, the IoT integration successfully simulates and validates the feasibility of human-in-the-loop task injection within a realistic, stochastic scenario.

Sensors doi: 10.3390/s26123896

Authors: Welker Facchini Nogueira Miguel Angelo de Carvalho Michalski Arthur Henrique de Andrade Melani Luiz David Ricarte de Souza Custodio Demetrio Cornilios Zachariadis Gilberto Francisco Martha de Souza

Anomaly detection models trained exclusively on healthy data are widely used in wind turbine condition monitoring because failure data are scarce, heterogeneous, and often unavailable. However, these models produce anomaly indicators that are sensitive not only to fault-related degradation but also to normal operational variability, transient disturbances, and changes in loading conditions. As a result, the practical behavior of an alarm system depends not only on the anomaly detection model but also on the decision rule used to activate and maintain alarm states. This study presents a decision-oriented evaluation of persistence-based alarm confirmation in wind turbine anomaly detection. Four representative techniques are analyzed within a unified framework: Isolation Forest, One-Class Support Vector Machine, Referenced Moving Window Principal Component Analysis using Q-statistic and percentage component weight indicators, and Autoencoder-based reconstruction error. The evaluation combines controlled OpenFAST simulations of rotor unbalance under different severity and noise conditions with an industrial SCADA case study involving a documented main bearing fault. Results show that temporal persistence strongly shapes alarm outcomes across methods and datasets. Low persistence values favor early detection but promote alarms from isolated threshold exceedances, whereas moderate persistence substantially reduces false positives while preserving detection capability in severe and well-observable faults. Excessive persistence increases detection latency and missed detections, particularly for weak, intermittent, or slowly evolving fault signatures. These findings indicate that persistence-based alarm confirmation should be treated as an explicit decision-level configuration variable, rather than as a fixed post-processing or alarm-state heuristic, when designing anomaly detection systems for wind turbine condition monitoring.

Sensors doi: 10.3390/s26123895

Authors: Enrique Albalate-Prieto Noelia Vallez José Luis Espinosa-Aranda Aubrey Dunne Raúl Barba-Rojas

Recent advances in satellite technology have significantly progressed, yet acquiring high-quality images with meaningful labels for Earth observation missions remains a costly and time-intensive process. Furthermore, captured scenes frequently exhibit defects such as misaligned color channels, extensive cloud cover, or repetitive patterns in similar environments. Fortunately, the evolution of generative artificial intelligence offers a solution by enabling the creation of realistic synthetic scenes, simulating the characteristics of any targeted imager, and thereby mitigating the scarcity of authentic data. This paper demonstrates the feasibility of transferring knowledge from specialized AI-generated datasets to Earth observation missions. Leveraging a novel dataset of Spanish map tiles, Pix2Pix, CUT, and ControlNet models were implemented to synthesize satellite imagery. To analyze structural and topological generalizability, identical U-Net instances were trained on the resulting collections for building, road, and water segmentation tasks, and subsequently tested on independent authentic imagery. The results reveal a clear decoupling between visual realism and functional utility. Incorporating synthetic samples into hybridized training datasets successfully surpassed the limitations of using real data alone, increasing maximum Dice scores by 0.9% (to 54.1% for buildings), 2.3% (to 38.6% for roads), and 4.1% (to 46.5% for waterbodies). This systematic validation establishes structural-guided synthetic data augmentation as a robust, adaptable strategy for Earth observation applications across diverse sensors and geometric objectives.

Sensors doi: 10.3390/s26123894

Authors: Kai-Chih Lin Rong-Jong Wai Hung-Yu Chang Chien

Falls among older adults remain a major public health concern, yet scalable and interpretable sensing approaches for functional fall-risk stratification remain limited. This study presents a lightweight contactless framework for five-level Short Physical Performance Battery (SPPB)-aligned functional fall-risk stratification using monocular RGB video. A total of 688 community-dwelling older adults completed SPPB-aligned assessments, including balance, five-times sit-to-stand, and 3 m gait tasks. Because prospective fall-event outcomes were unavailable, supervised labels were constructed from a pre-specified SPPB-aligned functional risk index rather than observed future falls. BlazePose-based two-dimensional keypoints were extracted, normalized using pelvis-centered and height-scaled transformations, and represented as temporal skeletal trajectories. Biomechanical descriptors were fused with embeddings from the proposed Temporal Convolutional Artificial Intelligence Fall-Risk Network (TCAI-FallNet). Participant-level data partitioning was used to reduce data leakage. TCAI-FallNet achieved a macro-averaged area under the curve of 0.91 and an overall accuracy of 81.3%. The trained model had a footprint under 3 MB, and TCN inference latency was below 20 ms per sequence under workstation-based evaluation. These findings suggest that TCAI-FallNet may support contactless SPPB-aligned functional mobility risk stratification, while prospective fall-event validation remains necessary.

Sensors doi: 10.3390/s26123893

Authors: Yunlong Bian Junhai Cao Chengming He Haidong Du Zhenwei Wang Xiaofeng Yue

The increasing diversity and complexity of modern mission scenarios have led to growing heterogeneity among nodes in mobile ad hoc networks: node functions, onboard devices, and operational parameters are becoming more diverse, and inter-node links are correspondingly no longer homogeneous. Such networks, termed heterogeneous functional chain networks, orchestrate nodes with distinct functions into multiple functional chains that cooperate to accomplish the overall mission. Accordingly, the evaluation of connectivity reliability in these networks has shifted from a topology-oriented paradigm to a functional structure-oriented one. This paper investigates the impact of dynamic reconfiguration mechanisms on the connectivity reliability of heterogeneous functional chain networks, accounting for node failures, node mobility, and link reliability. A Dynamic Reconfiguration Scheme (DRS) is designed based on the principles of minimum movement and minimum-ordinal decision node, and a suite of evaluation metrics—including normalized connectivity reliability, network quality, and connectivity reliability—is proposed together with a Monte Carlo simulation algorithm. The proposed approach is validated via MATLAB simulations involving 210 heterogeneous nodes organized into 70 functional chains. Results demonstrate that dynamic reconfiguration increases the terminal number of functional chains by 170.83% (from 12.10 ± 0.673 to 32.77 ± 2.241), improves normalized connectivity reliability by 170.73% (from 0.1729 ± 0.010 to 0.4681 ± 0.032), and enhances network quality by 82.96%. The connectivity reliability is further shown to evolve through three distinct temporal stages: an initial stable period where functional chains remain largely intact, a mid-stage fluctuation period characterized by iterative destruction–reconfiguration dynamics, and a late-stage degradation period triggered by candidate node pool depletion.

Sensors doi: 10.3390/s26123892

Authors: Zhe Wei Rui Wang Rong Dai Haibo Xu Huan Zhang Yurong Zou

Crack detection in mobile inspection scenarios is constrained by both the extremely slender geometry of crack targets and the real-time inference requirements on edge devices, which expose systematic limitations of general-purpose object detectors. This paper proposes YOLO-Crack, a closed-loop solution that couples geometry-statistics-driven module design with end-to-end edge deployment validation. On the algorithmic side, we first quantify crack geometric properties and then introduce (i) a crack-aware cross-dimensional fusion attention (CFCA) module to strengthen feature representations, (ii) a dual-path feature enhancement module (DFEM) to preserve fine details during upsampling, and (iii) an empirical smooth quality window adjustment with shape consistency regularization to stabilize bounding-box regression for slender cracks. Experiments on the Crack500 dataset show that YOLO-Crack achieves 78.8% precision, 51.4% recall, and 65.7% mAP@0.5, improving over the YOLOv11n baseline by 4.2, 1.7, and 2.9 percentage points, respectively. On the engineering side, we deploy YOLO-Crack on a Jetson Orin NX mobile robot platform and evaluate it in a real ROS pipeline; the measured end-to-end throughput reaches 25.5 FPS, meeting real-time video processing requirements. The proposed framework provides a practical reference workflow for edge vision tasks, from geometry analysis to engineering verification.

Sensors doi: 10.3390/s26123891

Authors: Kirill Nefjodov Mahmoud Ibrahim Anton Rassõlkin

Wireless power transfer (WPT) systems for electric vehicles (EVs) are increasingly being studied in the MHz range to increase power density and reduce the size of passive components. However, operation at higher frequencies significantly changes electromagnetic interference (EMI) behaavior. Fast switching in SiC- and GaN-based inverters, high-Q resonant operation, and frequency-dependent parasitic capacitances create conductive, capacitive, and magnetic interference mechanisms that are less significant in conventional kHz-range systems. Although many existing studies focus on power-transfer efficiency and converter optimization, EMI mechanisms in MHz-range EV WPT systems remain insufficiently systematized from a system-level electromagnetic perspective. This paper presents a state-of-the-art review of EMI generation mechanisms and system-level effects in high-frequency WPT systems for electric vehicles. The review considers the main interference sources and coupling paths, including switching-induced common-mode currents, resonant amplification of current and voltage stress, capacitive coupling between the coupler and nearby conductive structures, and magnetic-field redistribution caused by coil misalignment. Special attention is given to the transition from lumped-element assumptions to more distributed electromagnetic behavior at higher frequencies. The review also discusses the possible impact of these mechanisms on vehicle electronic subsystems and highlights the need for frequency-aware electromagnetic design, integrated modeling, and more rigorous EMC assessment for reliable MHz-range wireless EV charging systems.

Sensors doi: 10.3390/s26123890

Authors: Erandhi M. Liyanage Chun-Hung Lee Wen-Yen Chang Andrew An-Zhe Lee Guan-Hsiung Liaw Wu-Chuan Yang Yu-Hsin Liu Kun-Chan Lan Sai Ho Ling

Extrapyramidal symptoms (EPS) are motor side effects commonly induced by antipsychotic medications and can lead to measurable changes in handwriting patterns. These symptoms affect both the spatial and temporal characteristics of writing, including stroke thickness, direction and the rate of directional change. To model these complex variations, we propose a novel Liquid Generative Adversarial Network (LiquidGAN), which combines the adaptive dynamics of liquid neural networks with the data generation capability of GANs. Handwriting data were collected from 94 patients with confirmed EPS and 30 healthy controls using Archimedean spiral patterns drawn with both hands. A total of 211 images were processed for both binary and multiclass classification using a pretrained ResNet50 model. The pretrained ResNet50 achieved 92% accuracy and 97% precision in the binary classification task; however, its performance dropped significantly to 57% accuracy in multiclass classification, indicating limited capability in capturing fine-grained EPS severity variations. In contrast, the proposed LiquidGAN demonstrated excellent performance in the binary classification task, achieving 97% accuracy and 98% precision. More importantly, LiquidGAN substantially outperformed the baseline in the more challenging multiclass setting, achieving 70% accuracy and precision across four classes (mild, moderate, severe, and control). This shows that the diverse dataset from the liquidGAN significantly improves the HOG-ANN classification and effectively captures complex and subtle handwriting variations associated with different EPS severity levels that conventional models such as ResNet50 fail to distinguish. In addition, LiquidGAN generated diverse and realistic synthetic handwriting samples, yielding improved Fréchet Inception Distance (FID), precision, and recall compared with style GAN. These findings demonstrate that handwriting biomarkers, when analyzed through dynamic generative learning, offer an effective and non-invasive approach for monitoring extrapyramidal side effects in clinical settings.

Sensors doi: 10.3390/s26123889

Authors: Qi Liu Weijun Liu Hongyou Bian Fei Xing

Selective laser melting (SLM) equipment operates as a complex cyber–physical system, wherein strong implicit coupling among internal subsystems presents significant challenges for condition monitoring and fault diagnosis. Existing deep learning methods often suffer from feature submersion when processing multi-source heterogeneous data and lack the capability for system-level topological causal inference. To address these issues, we propose a multisystem coupling modeling and anomaly traceability method based on a decoupled graph attention network (ST-DBGAE). Independent local spatiotemporal feature alignment modules are constructed to map heterogeneous sensory data into a unified latent space. This eliminates dimensional discrepancies while strictly maintaining the feature independence of underlying hardware subsystems, such as optical and gas circuits. A dynamic graph attention mechanism with sparse priors is subsequently introduced to adaptively capture time-varying coupling weights triggered by implicit interactions (e.g., thermal fluids), bypassing the need for predefined rigid physical connections. Furthermore, a dual-branch two-stage decoupled optimization architecture is designed. By blocking the cross-interference of global backpropagation, this architecture outputs a continuous equipment health index (HI) based on reconstruction errors and employs a topological difference matrix inference mechanism to reversely anchor the root-cause nodes responsible for cross-system cascading degradation. Experimental results based on over 310,000 real operational monitoring records from industrial SLM equipment demonstrate that the proposed model achieves a comprehensive diagnostic Macro-F1 score of 96.5% across eight operating states. The single-class detection rates (ACCs) of specific underlying anomalies are significantly improved. This method not only enables high-precision equipment health warnings but also provides a physically interpretable microscopic fault propagation mapping for predictive maintenance.

Sensors doi: 10.3390/s26123888

Authors: Xiang Zhou Jiawei Sun Jiannan Zhao Feng Shuang

The Vortex Ring State (VRS) poses a catastrophic aerodynamic threat to coaxial dual-rotor unmanned aerial vehicles (UAVs). Traditional reactive detection mechanisms provide insufficient altitude for recovery, while existing data-driven diagnostics are severely bottlenecked by data leakage, extreme class imbalance, and a lack of physical interpretability. To bridge these gaps, this paper proposes a physics-informed multimodal deep learning framework that transitions from post-occurrence detection to proactive early warning. We establish a 1.5 s precursor window—creating a three-class ordinal state space—to provide the flight control system with critical intervention time for differential rotor recovery. We developed a novel ViT-LSTM architecture (MTSF-Net) to fuse continuous seven-channel onboard-recorded data (comprising three-axis acceleration, three-axis angular velocity, and barometric vertical velocity), which are subsequently transformed into Continuous Wavelet Transform (CWT) spectrograms. To ensure real-time unidirectional inference while preserving absolute physical vibration scales across heterogeneous sensors, a Calibrated Benchmark Normalization (CBN) strategy is introduced. Furthermore, a Hybrid Ordinal Loss is proposed to mitigate the extreme sample imbalance (<0.5%) of the precursor state by penalizing asymmetric aerodynamic degradation. Evaluated under a strict sortie-based isolation protocol, the proposed system achieves an exceptional test accuracy of 98.26% and an unprecedented precursor recall of 100%. Notably, it completely eliminates fatal missed detections (VRS predicted as Normal) and false-positive VRS predictions triggered by precursor states. Finally, Gradient-weighted Class Activation Mapping (Grad-CAM) is utilized to verify that the multimodal sensor processing pipeline successfully anchors onto authentic physical vibration frequencies rather than artifactual noise, laying a rigorous, interpretable foundation for intelligent aviation safety systems.

Sensors doi: 10.3390/s26123886

Authors: Wajiha Zaheer Chukwunonso Henry Nwokoye Seyedeh Negar Afrasiabi Khalil El-Khatib Li Yang

The increasing deployment of IoT-enabled electric-vehicle charging networks has created a rapidly evolving cyber–physical environment in which security mechanisms must operate amid ever-changing data patterns and resource constraints. In these environments, static Machine Learning (ML) pipelines are often insufficient because they struggle to adapt to concept drift issues, emerging attacks, and real-time operational requirements. We analyzed cybersecurity vulnerabilities, challenges of conventional ML approaches, and the possibilities of AI-powered, adaptive security measures. This paper examines Online AutoML and its advantages, including automated adaptation to streaming data, reduced human intervention, and privacy-preserving, resource-aware learning. Furthermore, this paper discusses adversarial attacks and defences in Online AutoML systems, highlighting the need for frameworks that jointly address concept drift, scalability, privacy, and adversarial threats. Finally, this study emphasizes the importance of establishing comprehensive public benchmarks for Online AutoML research.

Sensors doi: 10.3390/s26123887

Authors: Abdelmonem M. Ibrahim Doaa A. Fakhry Fares Al-Shargie

Feature selection is crucial for high-dimensional sensor and biomedical data because it reduces redundancy, improves generalization, and supports interpretable biomarker discovery. In this study, we propose a Binary Chaos-Enhanced Newton–Raphson-Based Optimizer (BCNRBO) for wrapper-based feature selection. The method integrates chaotic search dynamics, a Hamming-distance-based Dynamic Potential mechanism, and a new binary transfer function to enhance exploration and prevent premature convergence. BCNRBO was evaluated on 26 benchmark datasets using a variety of classifiers, including K-nearest neighbor (KNN), decision tree (DT), Naive Bayes (NB), and Support Vector Machine (SVM) classifiers. The proposed method consistently achieved competitive or superior classification performance while selecting fewer features than competing binary metaheuristic methods. In particular, BCNRBO consistently achieved the best feature reduction performance across all classifiers and secured the top Friedman rankings for DT, NB, and SVM, demonstrating its overall effectiveness. Statistical tests confirmed significant improvements over competing methods in most pairwise comparisons. These results suggest that BCNRBO is a promising feature selection strategy for sensor-derived biomedical and neurorehabilitation data, where compact and reliable digital biomarkers are needed.

Sensors doi: 10.3390/s26123885

Authors: Chuting Wang Ruijuan Xue Xuesong Mei Zuguang Huang

Heterogeneous fault information and a lack of real-time synchronization in CNC machine tools hinder effective Prognostics and Health Management (PHM). This paper designs and implements a digital twin-driven PHM framework for machine tools that integrates a unified machine-tool fault information dictionary and a mechanism-data dual-driven diagnostic model (ResNet-TCN). A cyber–physical platform was developed using OPC UA and RESTful APIs to ensure real-time data synchronization. Experiments on the PHM 2010 dataset demonstrate that the proposed ResNet-TCN model achieves a root mean square error (RMSE) of 5.46 μm for tool wear prediction. Its performance surpasses that of traditional LSTM models, and the proposed framework effectively eliminates information silos, providing a responsive, scalable and accurate PHM solution for smart manufacturing.

Sensors doi: 10.3390/s26123884

Authors: Wentao Cheng Tie Liu Yue Huang Weiyi Mao Anming Bao Yousef A. Al-Masnay Peng Du Zhiyong Zhang Ying Liu

The Gaizi River Basin, an alpine region in China crossed by the Karakoram Highway, is highly prone to glacier-related debris flows (GDF). Accurate debris flow susceptibility assessment in this high-altitude area remains challenging due to complex terrain, active tectonics, and dynamic glacial processes. This study develops a hybrid model integrating statistical methods and machine learning-based anomaly detection for debris flow susceptibility mapping. To address data noise, certainty factor (CF) distributions of debris flow predisposing factors (DFPFs) were derived via Locally Weighted Scatterplot Smoothing (LOWESS). The strength of the association between DFPFs and GDF susceptibility was evaluated using the mean residual between the raw and LOWESS-smoothed CF values. Multiple anomaly detection algorithms, including distance-based (L2 Norm), density-based (One-Class SVM), ensemble (Isolation Forest, RandNet), and GAN-based (WBiGAN-GP) methods, were tested on raw and CF-transformed data, using only the GDF inventory as the label. The CF-WBiGAN-GP model delivers the most balanced performance, excelling at identifying both high- and low-susceptibility zones. Results show that distance to stream, slope, and the topographic roughness and wetness indices are strongly associated with GDF susceptibility. Distance to glacier and precipitation appear less informative for direct susceptibility inference under our specific dataset and analytical setup.

Sensors doi: 10.3390/s26123883

Authors: Juan Bonastre-Egea Andrés Bueno-Crespo Juan Morales-García

Air quality forecasting and environmental health research at urban and regional scales depend on the combination of measurements from heterogeneous sensor networks, yet the construction of integrated multi-source datasets is rarely described or released as a self-contained deliverable. This paper presents an open dataset that combines four sensor-derived sources covering the whole of Spain over the period from 2022 to 2024: hourly air quality observations from the 588 stations of the national network operated by the Ministerio para la Transición Ecológica y el Reto Demográfico (MITECO), daily meteorological records from the Agencia Estatal de Meteorología (AEMET), daily mobility indicators derived from anonymised mobile telephony events published by the Ministerio de Transportes y Movilidad Sostenible (MITMA) at the municipality level, and a calendar of national and Autonomous Community public holidays. The processing pipeline harmonises sources that differ in temporal resolution, spatial codification and quality regime into a tidy hourly table indexed by station and timestamp, with a fixed feature schema of 56 variables per record. Air quality stations are paired with their nearest AEMET station through a three-tier distance rule, and the daily exogenous features are aligned to the air quality time axis through a two-variant temporal-alignment scheme (lag-and-expand to the hourly grid for the hourly release, same-calendar-day join for the daily release). A complementary daily resolution variant of the dataset is also released, with 72 columns and the same feature schema except for the air quality block, which is aggregated to daily mean, minimum and maximum. The integrated dataset contains approximately 15 million hourly records across the 588 stations and is released on Zenodo (DOI 10.5281/zenodo.20196221) under a Creative Commons Attribution 4.0 International (CC BY 4.0) licence. It is intended as a substrate for research on air quality forecasting, environmental epidemiology and multi-source data fusion at the nationwide scale.

Sensors doi: 10.3390/s26123882

Authors: Ruiqi Cao Yanqun Tang Caiqin Li Sitong Li Yicong Su Xinyan Ma Wei Li Miao Zhang

With the emergence of mobile sixth-generation (6G) integrated sensing and communication (ISAC) scenarios, conventional multicarrier waveforms face challenges in maintaining reliable communication and robust physical-layer security. In this paper, we propose a multi-layer encryption multiple-input multiple-output (MIMO) affine frequency division multiplexing (AFDM) with index modulation (IM) scheme, which exploits the inherent flexibility of the AFDM modulation parameter c2 and subcarrier IM to construct a multi-dimensional physical-layer security mechanism. To enable sensing and exploit MIMO spatial diversity, a unified downlink MIMO configuration is adopted, where sensing and communication share the same transmit waveform, receive array, and physical propagation environment. The proposed configuration enables multi-dimensional parameter estimation, including delay, Doppler, and angle. The obtained sensing information further assists beamforming design, channel reconstruction, and signal equalization. Furthermore, the base station and user equipment share synchronized secret keys, and a unified detection framework is developed to balance computational complexity and detection accuracy while remaining compatible with the multi-dimensional encryption structure of the MIMO-AFDM-IM system. Simulation results verify the effectiveness of the proposed scheme in mobile scenarios, demonstrating enhanced multi-dimensional sensing accuracy, improved resistance to eavesdropping, and superior communication reliability and energy efficiency (EE).

Sensors doi: 10.3390/s26123881

Authors: Hyunsu Cho Euijin Kim Wonsuk Choi

Multirotor unmanned aerial vehicles (UAVs) rely on electronic speed controllers (ESCs) that decode motor commands from pulse-width modulation (PWM) signals, making the flight-controller-to-ESC command path a physical-layer attack surface for intentional electromagnetic interference (IEMI). This paper presents a mechanism-based analysis of IEMI attacks that induce motor stoppage in UAV brushless DC motor controllers. We develop a timing-error model in which a sinusoidal disturbance on the PWM line shifts the detected edge instants and drives the decoded pulse width into stop-equivalent regimes, and we show that the disturbance reaching the ESC’s thresholding node is shaped by a frequency-selective cascade of the PWM cable’s coupling response and the ESC’s input-path transfer function. We experimentally characterize this model on five commercial ESCs through conducted and radiated injection. The measured thresholds differ by more than an order of magnitude across ESCs and are reordered between frequency bands and injection modes; comparing conducted and radiated results allows us to attribute these differences primarily to the cable coupling response and reveals cases where it either hides or amplifies an ESC’s susceptibility. The susceptible frequency also shifts with PWM cable length in qualitative agreement with transmission-line resonance, confirming that observed radiated susceptibility reflects the joint design of ESC and cable rather than a single intrinsic property. The cable lengths examined here (45–125 cm) are longer than those of compact multirotors and were chosen to place resonances within our antenna’s band; we discuss the implications of this choice and identify shorter, deployment-realistic cables as a priority for future work.

Sensors doi: 10.3390/s26123880

Authors: Jie Xiong Qiong Lou Fang Lu

Accurate tooth segmentation plays a crucial role in computer-aided dental diagnosis and treatment planning, particularly in applications such as tooth detection, lesion localization, orthodontic analysis, and implant surgery. However, panoramic dental X-ray images often suffer from tooth adhesion, low contrast, and blurred boundaries, making precise delineation difficult and potentially compromising downstream clinical analysis. To address these challenges, we propose a boundary-aware segmentation framework, termed Boundary-Aware ResU-Net (BA-ResUNet), which is built upon a ResU-Net backbone and enhanced with Coordinate Attention (CA) and explicit boundary modeling mechanisms. Specifically, CA modules are introduced into the encoder to improve spatial representation and positional awareness. In addition, a Boundary Extraction Module (BEM) is designed to capture boundary priors from shallow and deep features, while a Boundary Injection Module (BIM) progressively incorporates these cues into the decoder through foreground enhancement and background suppression. This design enables the network to better preserve inter-tooth gaps and improve boundary delineation. Experiments on the MICCAI STS-2D dental dataset demonstrate that the proposed method achieves superior performance in terms of Dice and IoU compared with representative existing methods. Ablation and qualitative analyses further show that CA and BEM/BIM play synergistic roles in improving regional overlap and boundary localization, particularly in challenging cases involving adhesion, low contrast, and indistinct contours. These results indicate that the proposed framework provides a reliable and effective solution for panoramic tooth segmentation and has promising potential for computer-aided dental applications.

Sensors doi: 10.3390/s26123879

Authors: Zhiyang Zhang Hongji Xing Ximing Yu Xiaogang Tang

Passive reception of spaceborne synthetic aperture radar (SAR) signals is of great significance for acquiring target characteristics and identifying SAR operating states. With the rapidly growing demand for high-quality SAR signal reception, signal-receiving arrays are prone to beam performance deterioration and difficulty in beamforming under wide-angle scanning conditions. Traditional uniform arrays fail to meet practical engineering requirements and cannot balance multiple conflicting performance indicators. To address the above technical bottlenecks, this paper proposes a design method of a non-uniform planar receiving array based on the MOEA/D-HPSO algorithm. Taking maximum sidelobe level (MSL), array gain (G), and beamwidth (BW) as core performance indicators, a multi-objective optimization model of SAR signal-receiving array for wide-angle scanning is established. This method integrates the multi-objective decomposition strategy and hybrid genetic particle swarm optimization mechanism, decomposes complex multi-objective problems into several scalar subproblems, obtains uniformly distributed Pareto fronts, and effectively improves the diversity of solution sets. Simulation experimental results show that the proposed algorithm is superior to traditional mainstream algorithms such as NSGA-II and MOEA/D-DE in terms of convergence accuracy, solution set distribution, and various performance indicators. Typical array design examples verify that the proposed method can adapt to various engineering application scenarios and provide technical support for spaceborne SAR signal reception and spectrum management.

Sensors doi: 10.3390/s26123878

Authors: Yukun Wang Chenhao Sun Ruifeng Liao Lijun Luo Jiefeng Duan

Offshore wind turbine blades are chronically exposed to complex marine environments with high humidity, salt spray, strong wind, waves, and intense radiation. Under such conditions, blade defects often exhibit small sizes, weak visual features, and heterogeneous visible infrared manifestations. Conventional single-sensor monitoring and empirically weighted fusion methods are insufficient for reliable defect diagnosis and risk grading. To address this problem, this paper proposes a cross-sensor consistency-guided dual-spectrum fusion framework, termed CG-DSF, for offshore wind turbine blade defect diagnosis and risk assessment. First, visible-light images and infrared thermal images are acquired by UAV-mounted imaging sensors, and sensor-specific branches are constructed to extract surface structural features and thermal anomaly responses. Second, visible and infrared features are aligned at the feature token level, and cross-sensor evidence is evaluated for spatial consistency, diagnostic semantic consistency, and anomaly consistency. A reliability-aware fusion strategy is then used to suppress low-quality or conflicting observations and construct a unified defect representation. Finally, a series of representative simulation case studies are carried out to comprehensively assess the overall performance and practical applicability of the constructed model. Experimental results reveal that the proposed framework possesses evident advantages in blade defect identification for offshore wind turbines, offering a feasible solution for advancing proactive and intelligent condition-based operation and maintenance of offshore wind assets in complex marine environments.

Sensors doi: 10.3390/s26123877

Authors: Ahmet Turan Tasdemir Erkan Caner Ozkat Gozde Yalcin Ozkat Fatih Gul

Black tea quality is governed by aroma chemistry: terpene alcohols (linalool, geraniol, nerolidol), methyl salicylate, and short-chain aldehydes whose abundance and release kinetics from the polyphenol-rich leaf matrix shape perceived grade. Grade information lies not only in the average headspace concentration but in the temporal shape of volatile organic compound (VOC) release under controlled heating. Conventional electronic noses obscure this signal: they rely on multi-sensor arrays, compress each response into summary statistics, and report accuracy only at the level of individual measurements. Whether a single low-cost metal–oxide–semiconductor (MOS) gas sensor can recover grade-defining aroma chemistry, and whether waveform-level modeling can exploit it, was therefore investigated. A portable electronic nose built around a Bosch BME688 sensor recorded 90 time series, each comprising four directly measured channels (temperature, humidity, pressure, gas sensor resistance) and a derived indoor-air-quality (IAQ) proxy computed from them by the on-chip BSEC library, from 16 commercial Turkish black teas across three quality grades. Two representations were compared on the same data: a feature-based pipeline reducing 25 statistical descriptors to seven principal components for six classifiers (best F1-macro = 0.624, MLP), and a raw-waveform Multi-Scale 1D-CNN with Squeeze–Excitation and temporal self-attention (MS-CNN-Attention). Under product-grouped cross-validation, the deep model reached F1-macro = 0.811 (+30%) and graded 14 of 16 products correctly by majority vote, against 11 of 16 for the MLP, with the largest gain in the medium grade (F1: 0.52 → 0.79), where summary-statistic compression destroys the release-kinetic signal. The contributions are threefold: one programmable MOS sensor operated as a thermal-desorption profiler rather than a sensor array; a direct comparison of feature-based classical learning against raw-waveform deep learning on the same small, non-normally distributed dataset; and a product-level decision-consistency metric suited to batch screening. Pairing a low-cost MOS sensor with waveform-level modeling offers a rapid, non-destructive route to aroma-chemistry-based tea quality screening.

Sensors doi: 10.3390/s26123876

Authors: Rafał Gasz Mateusz Pluskota Krzysztof Schwierz

Integration of LiDAR and thermal sensing has become increasingly important in robotics, infrastructure diagnostics, environmental monitoring, and autonomous perception systems. LiDAR sensors provide accurate three-dimensional geometric information but do not directly capture thermal properties of observed objects, whereas thermal cameras provide temperature distributions without explicit spatial structure. Fusion of both sensing modalities enables thermally augmented 3D scene reconstruction and spatial localization of temperature anomalies. This paper presents a practical LiDAR–thermal fusion framework for three-dimensional localization of heat sources using an Ouster OS1 LiDAR sensor and a FLIR A70 thermal camera. The proposed framework includes intrinsic thermal-camera calibration, extrinsic LiDAR–thermal calibration, multimodal data synchronization, projection of LiDAR points onto the thermal image plane, and assignment of temperature values to spatial points. Additionally, a dedicated thermally distinguishable calibration target is proposed to enable reliable multimodal feature extraction under low-contrast LWIR imaging conditions. The developed framework was experimentally validated using real radiometric thermal data and LiDAR point clouds acquired under laboratory conditions. Quantitative evaluation demonstrated reprojection errors below 1 pixel and a mean hottest-point localisation error of approximately 4.1 cm at a distance of 12.3 m. The results confirm that accurate spatial localisation of thermal anomalies can be achieved using a geometry-based multimodal fusion approach without relying on computationally expensive learning-based methods. The proposed framework emphasises practical deployment, deterministic calibration, and applicability in scenarios where limited training data or constrained computational resources make learning-based approaches difficult to apply. The proposed system may be applied to building energy diagnostics, industrial inspection, technical infrastructure monitoring, and robotic perception systems that require reliable spatial localisation of heat sources under real measurement conditions.

Sensors doi: 10.3390/s26123875

Authors: Jiong Guo Yue Wang Dan Wu Lili Qiu Zhibin Xu Junming Geng Yifei Wang Zihui Meng

A silk fibroin/cellulose inverse-opal photonic crystal composite with robust mechanical properties was fabricated by blending a silk fibroin solution with methylcellulose, utilizing a 3D poly(methyl methacrylate) (PMMA) photonic crystal array as a template, via sequential infiltration, curing, and etching processes. Leveraging the intrinsic water sensitivity of both silk fibroin and methylcellulose, the resulting composite exhibits exceptional moisture-sensing capabilities across gaseous, liquid, and solid phases. Specifically, for atmospheric humidity, the film delivers a distinct optical response to a relative humidity variation in merely 5%. In liquid systems, owing to the material’s excellent affinity for low-polarity organic solvents and the disruptive effect of highly polar solvents (e.g., water) on the photonic periodic structure, the structural color of the film can sensitively report trace water contents down to 0.025%. Furthermore, in solid matrices, the composite enables the precise detection of not only free water but also water of crystallization.

Sensors doi: 10.3390/s26123874

Authors: Amira J. Zaylaa Lama N. Yassine Silva Kourtian

Feature selection represents a critical step in developing accurate and interpretable models for early breast cancer detection. Despite extensive research in the field of mammographic image analysis, no consensus has yet been reached on the optimal feature subsets that distinguish normal from malignant tissues. To address this gap, the present study aims to identify the most discriminative and significant features through a comprehensive multi-criterion selection framework. The aim is to integrate, as new frameworks, different combinations of t-test, ANOVA, Mutual Information (MI), and Equal Grouping Methods (EGM) to rank 19 linear and nonlinear features extracted from mammographic images. The objective is to maximize feature relevance while minimizing redundancy and enhancing diagnostic and healthcare systems. Linear features were assessed alongside nonlinear descriptors. A framework combining t-test, ANOVA, and EGM, guided by MI relevance, was employed to balance feature contributions across categories. The experimental results demonstrated that hybrid feature selection significantly enhanced diagnostic accuracy using optimal linear and nonlinear attributes. The optimization results suggested using a hybrid of six linear and eight nonlinear features. Linear features were highly accurate for detecting cancer. Haralick entropy obtained the highest average accuracy and performance, 94.14% and 93.45%; followed by kurtosis, 93.49% and 92.59%; perimeter irregularity, 93.43% and 92.65%; skewness, 93.01% and 92.25%; and volume/area, 92.82% and 91.92%. Despite the reliable discriminative power of linear descriptors, their overall effectiveness in representing intricate tissue characteristics was limited. The comparison of statistical characteristics shows a distinct performance benefit of nonlinear descriptors over linear ones for detecting breast cancer. Nonlinear descriptors, however, showcased higher accuracy and performance, with an average accuracy of 97.81% in contrast to 94.43% for linear approaches. Local phase congruency achieved the top average accuracy and performance, 97.81% and 96.61%, respectively; succeeded by wavelet entropy, 97.62% and 96.42%; Laplacian spectrum features, 97.52% and 96.32%; nonlinear diffusion, 97.10% and 95.90%; and clustering coefficient, 96.70% and 95.50%; then Shannon, Tsallis, and Rényi entropies. The results indicate that statistically validated nonlinear characteristics significantly outperform linear ones across accuracy and performance measures. Their combination significantly improves the strength and discriminative power of computer-assisted breast cancer diagnostic systems, affirming their suitability for integration into sophisticated machine learning and deep learning models. The results also show that the new multi-criterion framework’s early detection performance surpassed that of the statistical and deep learning models explored, with an average of 98.6% accuracy, 98% sensitivity, 98.9% precision, and 98.4% F1 score of early detection of breast cancer. The incorporation of statistically validated nonlinear descriptors, particularly local phase congruency and wavelet entropy, improves the discriminative ability, robustness, and clinical understanding of breast cancer computer-assisted diagnostic systems. Overall, the proposed framework confirms that integrating hybrid features substantially enhances robustness and plays a pivotal role in computer-assisted breast cancer detection. These selected features may be fed to more advanced algorithms in the future, potentially yielding improved performance.

Sensors doi: 10.3390/s26123873

Authors: Enhao Cui Runshan Hu Weina Zhang Zihan Fei Chenyang Zhu

Cross-material continuous tool wear prediction is difficult because a model must preserve the physical wear scale, not only align high-dimensional sensor features. This limitation is critical in milling, where the target variable is the continuous flank wear width (VB) and material shift can distort the mapping from sensor response to wear magnitude. We address this problem by recasting cross-domain tool wear prediction as monotone wear-scale adaptation. We propose Multi-Physics Monotone Score Transport (MPMST), a monotone score transport framework that constructs a tool-wear-oriented score from sensor-derived candidate cues, transports the target-domain score onto the source-domain wear scale, and then predicts wear through isotonic regression. We also evaluate One-Physics Monotone Score Transport (OPMST), a force-only variant that uses the same score-transport pipeline with a restricted cue family. On Mondragon Unibertsitatea–Tool Condition Monitoring (MU-TCM) with two cross-material transfer tasks, the validation-driven MPMST configuration reduces mean absolute error by approximately 63% relative to Correlation Alignment (CORAL) and by approximately 31% relative to a physics-informed Gaussian process baseline. The results support monotone score construction and score transport as practical mechanisms for continuous tool wear prediction under domain shift, while also showing that MU-TCM is strongly force dominated.

Sensors doi: 10.3390/s26123872

Authors: Hee-Jin Jun Kang-Hyuk Lee Dulce Urueta Tapia Jerel P. Calzo Heather L. Corliss

LGBTQ+ individuals experience disparities in cardiovascular health, but little is known about how daily minority and general stress affect physiological and psychological responses in real-world settings. Twenty LGBTQ+ young adults aged 18–27 completed a 14-day exploratory pilot study using ecological momentary assessment (EMA) with four daily smartphone surveys and continuous smartwatch-based sensor monitoring. This study is among the first to combine EMA with wearable sensor data to capture autonomic stress responses to minority stressors in naturalistic settings. Outcomes included a physiological stress score derived from heart rate variability during the 60 min before each EMA completion, as well as positive and negative affect (PA and NA). Four stress measures, Everyday Discrimination Scale (EDS), Sexual Orientation Microaggression Inventory Short Form (SOMI-SF), EMA of stressful events (EMA-SE), and current perceived stress (CPS), and a combined variable (COMB) were examined. In mixed-effects within-person models, all stress measures showed trends in the expected direction, with higher physiological stress scores, lower PA, and higher NA, though these varied in magnitude and statistical significance. SOMI-SF showed the strongest association with physiological stress, while general stress measures showed stronger associations with affect. These preliminary findings suggest that LGBTQ+-specific and general stressors may differentially engage physiological and psychological response systems; however, caution is warranted given the small sample size.

Sensors doi: 10.3390/s26123871

Authors: Yonghao Li Fan Wu Ping Ye Qingxuan Jia

Multi-agent perceptual map construction and long-term maintenance constitute an important paradigm for improving adaptability and real-world applicability. With the outstanding capability of 3D Gaussian Splatting in preserving fine-grained texture details, a number of 3DGS-based real-time mapping approaches have recently emerged. However, these methods often struggle to cope with complex dynamics in real-world environments and lack the generalization needed to scale to multi-agent systems. Existing solutions typically rely on direct parameter concatenation or locally confined optimization, which are unable to explicitly model cross-agent observation reliability under temporal asynchrony and dynamic inconsistency, and therefore tend to amplify conflicting updates rather than resolve them. To address these limitations, we propose DGOMapping, an online system for multi-agent dynamic perceptual mapping. DGOMapping leverages an uncertainty-coupled 4DGS scene representation and a collaborative interaction mechanism via Gaussian perception-score exchange, enabling both real-time 4DGS construction and long-term map memory adjustment. Experiments on multiple real-world datasets demonstrate that DGOMapping effectively suppresses dynamic interference and exploits multi-agent collaboration, achieving state-of-the-art performance in both tracking and reconstruction. The proposed system therefore provides a practical sensing-oriented solution for collaborative perception and real-time dynamic environment mapping.

Sensors doi: 10.3390/s26123870

Authors: Dian Li Peiji Chen Shunta Togo Hiroshi Yokoi Yinlai Jiang

Surface electromyography (sEMG) gesture recognition degrades across recording days under domain shift, increasing calibration burden for myoelectric interfaces. Many cross-day adaptation pipelines retrain the deployed recognizer or require labeled target-session data, which can be impractical in assistive-device settings where classifier versions may need to remain locked for traceability and regulatory compliance. We study unsupervised cross-day adaptation under two constraints: the task classifier remains frozen and holdout-day labels are not used when training the adaptor. We propose the Source-Reprojection Module (SRM), a plug-in front end that combines conditional adversarial feature learning with a residual signal-space projector guided by the frozen classifier’s gradients, identity regularization, and latent-space distribution matching, using labeled source days and unlabeled adaptation days only. On a multi-day protocol with four healthy participants (at least five calendar-day sessions per participant, split 3:1:1 into source, adaptation, and holdout domains) and three random seeds per participant (12 runs), mean holdout accuracy increases from 70.9% for the frozen classifier alone to 72.8% with SRM (+1.98±0.91 percentage points averaged across subjects). SRM outperforms the frozen baseline in 10 of 12 subject–seed runs. The gain is modest and the cohort is small, so the result supports proof-of-mechanism under the stated protocol rather than population-level clinical generalization.

Sensors doi: 10.3390/s26123869

Authors: Xiongfei Yin Sicheng Sun

Wavelength stability of distributed feedback (DFB) semiconductor lasers in dense wavelength division multiplexing (DWDM) systems hinges on sub-millikelvin temperature regulation, a task complicated by the nonlinear, multi-node dynamics of the thermoelectric cooler (TEC) and the purely reactive nature of conventional proportional–integral–derivative (PID) control. We present a physics-informed neural network (PINN) built around a residual correction architecture for hybrid feedforward–feedback TEC temperature control. Rather than penalizing physics-residual violations in the loss function, the architecture wires a simplified one-node thermal model directly into the network graph as a frozen baseline. A trainable branch then learns only the residual mismatch. Temporal lag features are appended to the input so that the network can reconstruct unmeasured internal thermal states from the cold-side temperature history, which proves essential for overcoming the partial-observability bottleneck inherent in multi-node TEC packages. Ablation experiments on a high-fidelity three-node TEC simulator show that all model variants (PINN, physics-feature-augmented NN, and pure NN) exceed R2 = 0.993 when trained on the full dataset, yet the PINN’s advantage becomes pronounced under data scarcity. At a 3% training budget, it reaches R2 = 0.966 versus 0.930 for the pure NN, implying an approximately 5.4× reduction in the data needed to reach a given accuracy target. In closed-loop validation, the PINN+PID hybrid settles 60% faster than standalone PID. Tracking RMSE drops by 69%, and peak disturbance deviation falls by 74%, across step, multi-setpoint, and current-perturbation scenarios. All results reported here are obtained in simulations. Experimental validation on physical DFB-TEC hardware is left to future work.

Sensors doi: 10.3390/s26123868

Authors: Xin Liang Huijiao Qiao Yanda Chen Jin Zhang

Building extraction from disaster scenes is critical for emergency response and post-disaster assessment. Unlike conventional static remote sensing imagery, multi-phase disaster imagery contains scenes spanning early, middle, and late disaster stages, where building morphology, class distribution, and boundary characteristics exhibit significant cross-phase heterogeneity. Such phase-dependent variations substantially increase the difficulty of stable semantic segmentation, particularly under complex damage conditions. To address these challenges, we propose MSS-MambaNet for building extraction from multi-phase disaster imagery. A multi-scale architecture is designed to overcome the limitations of single-scale scanning in Mamba, enabling more effective perception of diverse building morphologies. To enhance feature discrimination, a Dual-Domain Cross-Gated Fusion (DDCGF) module is introduced through complementary interactions between spatial and frequency-domain representations. In addition, a Pixel-Aware Dynamic Weighting (PADW) strategy is developed to adaptively emphasize imbalanced foreground pixels and ambiguous boundary regions, thereby improving segmentation consistency under complex disaster conditions. Extensive experiments demonstrate that MSS-MambaNet consistently outperforms state-of-the-art methods, achieving an average mIoU of 92.78% and mF1 of 96.25% with only 12.37 M parameters. These results indicate that the proposed method effectively handles the heterogeneity of multi-phase data, providing a stable and efficient solution for building extraction from multi-phase disaster imagery.

Sensors doi: 10.3390/s26123867

Authors: Maryam Arianpouya Benson Yang Peter Truong Simon J. Graham

Electric dipole antennas can be designed in a variety of geometries and applied across a wide range of configurations. Appropriately designed dipole antennas can provide deep tissue penetration and low radiofrequency (RF) power deposition in magnetic resonance imaging (MRI), making them attractive for applications requiring safe and effective RF transmission in deep regions. On clinical 3 T MRI systems, however, conventional dipoles are too large in size for practical imaging of the head. Inspired by telecommunications designs, the present work adapts meandered dipoles (where the conductor is folded to shorten the antenna) with the resonance frequency controlled through trace geometry. Additionally, multi-channel configurations are considered to improve RF power transmission. A straight dipole was progressively transformed into meandered geometries and characterized using benchtop measurements and electromagnetic simulations. Analyses evaluated frequency response, near-field behavior, power-flow directionality, and distributions of local tissue heating and transmitted RF magnetic field in multi-channel arrays. A four-channel parallel-transmit (pTx) prototype was also used to show the feasibility of dipole-based head imaging at 3 T. The present work demonstrates a practical implementation of compact, low-heating dipole arrays for head MRI, with potential for extension to ultra-high-field or multinuclear imaging.

Sensors doi: 10.3390/s26123866

Authors: Huijie Zheng Jing Chen Qi Lin

The rate control algorithm is designed for natural video by default in video-coding standards. However, computer-generated screen content video (SCV) is very different from natural video captured by a camera, with many different statistical characteristics, such as sharp edges, thin lines, and flat area. This will lead to a difference in the focus of the human visual system (HVS) when viewing on-screen content video. Especially in various sensor data visualization applications such as intelligent display terminals, industrial monitoring and human–computer interaction interfaces, screen content video carries key information collected and reconstructed by image sensors, vision sensors and multimodal sensors. Its edge structures and local details directly affect the interpretation accuracy and application reliability of sensor information. Therefore, it is crucial to investigate perceptual rate control methods that integrate both video content characteristics and human visual perception properties, which possesses substantial theoretical and practical significance. In this paper, we propose a perceptual rate control algorithm for screen content video based on just-noticeable distortion (JND) which is established on the edge profile reconstruction with tolerable variations. First of all, target bit rate allocation for the frame level and CTU level is based on a perceptual weight which is calculated on the JND factor and reconstruction edge character. Secondly, under the constraint of the JND model, an intra rate-distortion (RD) model is established under the constraint of the JND model. The similarity between reference frames and reconstructed frames is taken as feedback in this model. Finally, the proposed rate control algorithm (JND–perceptual rate control (PRC)) is integrated to the existing rate control framework in High-Efficiency Video Coding–Screen Content Coding (HEVC-SCC) for improving the coding efficiency. The experimental results show that the proposed algorithm achieves better bit control precision than the platform, as well as improves the R-D performance of screen content video. In particular, compared with the HEVC-SCC reference software, the coding performance is improved by 3.09 dB on average, the bit rate is saved by 26.51% on average, and the average bit rate mismatch is within 1.159%.

Sensors doi: 10.3390/s26123865

Authors: Jiazhi Dai Mario Rotea Nasser Kehtarnavaz

Fault detection is essential for the reliable operation of wind turbines. Traditional supervised methods for fault detection based on SCADA data face highly imbalanced datasets of normal and fault samples. This paper presents a two-stage detection method to address this limitation by integrating unsupervised anomaly detection or classification with supervised classification. In the first stage, the unsupervised classifier of OCSVM, together with two complementary anomaly scores, is used to flag deviations from normal operation or separate abnormal data samples from normal data samples. In the second stage, the supervised classifier of CNN is applied to the detected abnormal data samples to identify fault samples among only these samples, thus enhancing the discrimination capability between normal and abnormal conditions. Experiments on real-world SCADA data show that the introduced two-stage detection method noticeably improves fault detection compared to supervised methods, both in terms of accuracy and missed fault rates.

Sensors doi: 10.3390/s26123864

Authors: Xinyi Feng Shaochen Jiang Liejun Wang Beibei Gao

High-resolution remote sensing image segmentation is important for urban mapping but remains challenging because of spectral ambiguity, large scale variations, fragmented elongated structures, and background interference. This study aims to improve semantic segmentation in complex aerial scenes by combining local feature extraction, selective multi-scale fusion, and global sequence modeling. We propose DSD-Mamba, an asymmetric dual-stream architecture with a ResNet-18 encoder. The Dense-Sparse Pyramid Fusion Module aligns multi-level features and applies dual Top-k selective value aggregation for cross-scale response filtering and background-response suppression. This Top-k operation is used as a feature-selection mechanism and is not intended to reduce the theoretical memory footprint of dense attention. Scale-Aware Strip Attention refines skip connections through horizontal and vertical dependency modeling, and the Dual-Stream Context Decoder combines a Mamba-based global branch with a CNN-based local branch during upsampling. Experiments were conducted on UAVid, ISPRS Vaihingen, and ISPRS Potsdam under a single-model inference protocol without test-time augmentation. DSD-Mamba achieved mIoU scores of 73.4%, 85.2%, and 87.2%, respectively. Ablation experiments on Vaihingen showed that DSPFM, SASA, and DSCD improved performance over the baseline when evaluated in this setting, with the full model reaching the highest mIoU. The method improves segmentation accuracy under the tested protocols, although its higher FLOPs indicate an accuracy-oriented rather than lightweight design.

Sensors doi: 10.3390/s26123863

Authors: Ruiqin Bai Xiaopeng Zhang Xiaoyu Lv

As RFID systems become increasingly widespread, the limitations imposed by tag collisions on system performance are becoming more evident. Parallel decoding has attracted significant attention due to its ability to improve channel utilization and throughput. However, existing schemes often perform poorly when decoding weak signals. Several challenges remain, including the assumption of ideal channel conditions, difficulty in detecting tag state transitions, and the complexity of state cluster formations in the In-phase and Quadrature (IQ) domain. To address the above issues, this paper first experimentally verifies the ability of middle state points to segment tag states, and proposes a time-window-based pre-processing method to improve the density of state clusters in the IQ domain. Second, by leveraging the high vertical resolution of the reader, we propose an ideal magnitude calculation method and a matching strategy for combined state clusters under weak signal conditions. Finally, we propose MagMap, a parallel decoding scheme based on middle state points and magnitude extraction. Experimental results demonstrate that, under weak signal conditions, MagMap reduces the decoding BER (Bit Error Ratio) of received packets by more than 60% compared to the state-of-the-art.

Sensors doi: 10.3390/s26123862

Authors: Valentina Tirsu Andrei Dorogan Lilia Sava Larisa Dunai Alexandru Ilev Nelea Manin

Electromyographic (EMG) signals are increasingly used in wearable cyber–physical systems (CPS), where reliable movement recognition must be achieved under limited computational resources. In this study, we present a compact EMG processing framework that integrates signal acquisition, preprocessing, segmentation, and movement classification within a unified pipeline designed for embedded-oriented applications. The proposed approach combines a multi-channel EMG acquisition system with a lightweight one-dimensional convolutional neural network (1D CNN) developed according to TinyML principles, withprocessing input windows of size 32 × 3 and low computational complexity and memory requirements. Experimental evaluation was conducted on a dataset collected from 15 participants performing squat, walking, and running activities under realistic acquisition conditions. The proposed model achieved an accuracy of 0.9135, an F1-score of 0.9124, and a ROC AUC of approximately 0.96, demonstrating reliable classification performance. Following 8-bit quantization, the model size was reduced to approximately 2 KB, supporting deployment on resource-constrained embedded platforms. The results show that compact CNN architectures can effectively classify EMG-based movement patterns while maintaining a small computational footprint, providing a practical foundation for future wearable CPS and TinyML-enabled applications.

Sensors doi: 10.3390/s26123861

Authors: Haijun Lin Jiayi Liu Dongxu Jiang

Early depression screening is important for timely intervention, and electroencephalography (EEG) offers an objective and potentially portable sensing modality for computer-aided assessment. However, recognition from fixed three-channel frontal EEG remains difficult because of limited spatial information and incomplete modeling of temporal–frequency characteristics and temporal dependencies. This study proposes DEP-TFDualNet for acquisition-constrained frontal resting-state EEG. The framework integrates multi-scale convolution, dual-domain channel attention, temporal modeling derived from the independent recurrent neural network (IndRNN) architecture, and decision-stage fusion of deep representations with low-order statistical descriptors through a Kolmogorov–Arnold Network (KAN)-based nonlinear projection layer. Experiments were conducted on the publicly available three-channel frontal EEG subset of the MODMA dataset. After additional quality control, 48 subjects were retained (22 patients with major depressive disorder, 26 healthy controls). Under subject-wise stratified five-fold cross-validation, DEP-TFDualNet achieved 85.42% accuracy, 85.26% macro-F1, 81.82% sensitivity, 88.46% specificity, an AUC of 0.82, and a Brier score of 0.121. It achieved the best threshold-based subject-level performance and the lowest Brier score among the evaluated models. These results provide preliminary evidence that simplified frontal EEG sensing may support depression recognition in acquisition-constrained settings, although larger and external validation is still required.

Sensors doi: 10.3390/s26123860

Authors: Maximiliano Vélez Claudio Urrea

Cameras and IMUs on heavy mining trucks supply the visual signal that Advanced Driver Assistance Systems (ADASs) use in open-pit operations. Haul roads in a surface mine are unstructured and unmarked, so a perception model must be both accurate and fast. We address this with a video-based multitask pipeline for a mining Driver Support System (DSS): a single BiSeNetV1 network produces drivable-area segmentation and steering-direction classification in one forward pass. Training used only 100 frames sampled non-sequentially from in-cab recordings of a real open-pit mine; evaluation used two full onboard sequences. To exploit temporal redundancy without annotating video, we propose an Adaptive Clockwork (A-CW) inference scheme: the spatial path runs on every frame, while the context path is refreshed only on keyframes whose cadence is set by the classification output, the same signal shown to the driver as a steering hint. This classification-guided policy increases context updates on curved segments, where the scene changes more rapidly, and reduces them on straight sections, where semantic redundancy is higher. The selected A-CW configuration was evaluated on full temporal test sequences, including one route kept entirely outside the training source. On this unseen route, A-CW achieved 94.70% road-class IoU and 73.68% Top-1 Accuracy. GPU-only throughput increased from about 55 FPS with frame-by-frame inference to 168.01 FPS, and display-excluded end-to-end processing in the simulated ADAS pipeline remained at approximately 37.5 FPS.

Sensors doi: 10.3390/s26123859

Authors: Natalia Alegría Igor Peñalva Charles Pinto Adriana Merello

This study investigates the radiological characteristics of drinking water sources managed by the Bilbao Bizkaia Water Consortium (CABB). To this end, the radiological monitoring parameters established by current regulations, as well as those applied by other international organizations, are reviewed and analyzed. In addition, commercially available continuous monitoring equipment is assessed in terms of its suitability for drinking water applications. To identify optimal deployment locations, a comprehensive evaluation of CABB water infrastructure is conducted, with the aim of ensuring radiological safety across the Bizkaia region. Furthermore, an economic assessment is carried out to estimate the potential cost of water supply under abnormal contamination scenarios.

Sensors doi: 10.3390/s26123858

Authors: Marwah Albeladi Kamal Jambi Fathy E. Eassa Maher Khemakhem

Sixth-generation (6G) networks are expected to enable a new level of connectivity, with peak data rates reaching 1 Tbps and latencies below 0.1 ms, especially in large-scale Internet of Things (IoT) environments. Despite these advantages, the rapid increase in device density poses multiple challenges, most notably the growth in control plane signaling and the associated increase in energy consumption. These issues might significantly affect the scalability and efficiency of future networks if left unaddressed. We propose EASE-6G, an energy-aware Software-Defined Networking (SDN) framework that moves network operation from reactive to proactive and predictive, supporting ultra-dense conditions, where the number of connected devices may reach 106 devices per square kilometer. EASE-6G uses Proactive Flow Installation to reduce the need for instant decisions. Traffic is predicted using a Long Short-Term Memory (LSTM) model, while a signaling-aware Deep Q-Network (DQN) streamlines control, reducing unnecessary signaling while maintaining performance. Simulations in OMNeT++/Simu5G were performed to compare EASE-6G with Smart Fog Radio Access Network (SF-RAN) and Deep Q-Network-based Open Radio Access Network (DQN-ORAN). EASE-6G was found to reduce energy consumption by 36.8%, signaling overhead by 36.7%, and latency by 35.6%. The LSTM model achieved a Mean Absolute Percentage Error (MAPE) of 4.2%. The DQN agent showed improved stability, with 22% lower variance than the baseline. These results demonstrate that the proposed predictive SDN control mechanisms improve energy efficiency and reduce overhead, delivering a practical solution for the implementation of scalable, sustainable IoT in future 6G networks.

Sensors doi: 10.3390/s26123857

Authors: Liang Qi Jianjiang Zhou

Adaptive Sidelobe Cancellation (ASLC) is a core technology for modern radar systems to suppress active sidelobe jamming. From the perspective of disrupting the ASLC system’s ability to stably track the jamming direction, this paper proposes a distributed jamming method based on random phase perturbation. The method employs two spatially separated jamming sources that simultaneously transmit coherent signals. By actively applying controllable random jumps to the relative phase between the two sources, the equivalent wavefront direction of the synthesized signal at the radar receiver changes rapidly, forming a non-stationary jamming that destroys the null-tracking capability of ASLC. An analytical model of the ASLC cancellation ratio (CR) under random phase perturbation is established, with a focus on analyzing the effects of time synchronization accuracy and phase synchronization accuracy on jamming performance. Monte Carlo simulation results show that the proposed method can reduce the average ASLC CR from 26.80 dB to 20.29 dB (a decrease of 6.51 dB). Under identical conditions, this performance is comparable to asynchronous blinking jamming while requiring no precise timing matching, and outperforms multi-source saturation jamming in resource efficiency (two vs. four jammers). This study provides promising simulation-level evidence for the effectiveness of the proposed jamming method. The quantitative results and sensitivity analyses offer a simulation-level theoretical reference for parameter design of distributed cooperative jamming. Further validation in semi-physical simulations or field trials is necessary before claiming engineering readiness.

Sensors doi: 10.3390/s26123856

Authors: Qichen Yang Chun Yang Mingzhu Liu Nan Su Jingran Sun Jian Hou Yixue Fu Jin Wu Yu Wang Yuan Peng Jialei Bai Ying Liu Zunquan Zhao

Antigen detection provides rapid and convenient diagnosis of respiratory infections. This study develops an innovative dual-fluorescence aptasensing method based on polydopamine-functionalized MXene (PDA-MXene) for the simultaneous detection of spike protein and hemagglutinin protein. The method employs green- and red-emitting quantum dot (QD) probes as fluorescence reporters, and the PDA-MXene as an effective adsorption and separation substrate. Coupled with a centrifugation-assisted separation strategy, this design method reduces background interference and enhances detection reliability. The method demonstrates good analytical performance, with detection limits of 0.82 ng/mL for spike protein and 2.11 ng/mL for hemagglutinin protein in single-channel mode. The dual-channel mode enables reliable and simultaneous quantification of both target proteins with minimal spectral cross-talk. Furthermore, this method exhibits high specificity against interferents including ions, proteins, and toxins. Artificial saliva, chosen as real sample, is spiked with target proteins to investigate the practical applicability of the method, showing recovery rates for both target proteins between 100 and 114 sensing strategy is simple to operate and allows the detection of new targets by simply replacing the azide-modified aptamer lyophilized powder. It therefore holds promising application for the simultaneous detection of multiple proteins in point-of-care testing and health monitoring fields.

Sensors doi: 10.3390/s26123855

Authors: Onur Uysal Mehmet Emin Bakir Andres R. Perea Vedat Tumen Santiago A. Utsumi

Accelerometer sensors and artificial intelligence (AI) are reshaping automated behavior monitoring in precision livestock management, yet their joint deployment on extensive rangelands is constrained by energy and bandwidth budgets. Low-Power Long-Range Wide-Area Network (LoRaWAN) collars address these constraints by compressing the raw tri-axial signal on the device into a single scalar per reporting interval, the Motion Index (MI). This onboard compression preserves enough signal to separate active behaviors but discards the per-axis and frequency content that fine-grained classification typically relies on. On a dataset of 9222 labeled observations from 24 cows across four breeds, MI distinguishes walking from grazing reliably but fails to separate ruminating from resting; both correspond to a stationary animal and yield near-zero, statistically indistinguishable distributions. Earlier MI-only models reached only about 65% four-class accuracy, and ruminating was commonly merged into resting. We show that much of this loss can be recovered by treating the MI stream as a time series. Session-aware lag features, rolling statistics, and an autoregressive previous-behavior feature lift four-class macro-F1 from 0.647 to 0.94, with per-class F1 of 0.95 for ruminating and 0.92 for resting (and at least 0.92 for every behavior). In autonomous deployment the previous behavior must be predicted rather than observed; for this setting we add a Viterbi sequence-decoding step that combines the classifier’s per-step outputs with a learned behavior-transition model, recovering a substantial part of the ruminating signal from the activity stream alone while keeping walking and grazing reliable. The gain is consistent across seven classifiers and four genetically distinct breeds, indicating that it is driven by the features rather than by a specific model.

Sensors doi: 10.3390/s26123854

Authors: Miki Sirola Jaber McBreen Mohammad Raisi Esfarjani

The larger research problem is to get combustion engines more effective and flexible and reduce or even eliminate greenhouse gas emissions. Here we concentrate more on a smaller-scale and focused research problem about the significance of air feeding in engine operation. Therefore, the need for modeling a charge air system is obvious. The interaction and co-operation between the charge air systems and combustion engines is a central issue in this article. A literature review was carried out on related topics, and it reveals a research gap in this area. A simulation model of a charge air system based on first principles is developed. It is based on physical and systemic modeling, and it is constructed including control loops reducing and controlling the pressures in the charge air chain. The simulation models of this auxiliary system and engine are successfully combined, and functioning together is demonstrated. The composed models represent real research laboratory equipment in the University of Vaasa Energy Laboratory under construction. The research laboratory equipment and the whole research environment are described. Simulation scenarios are presented both with the charge air system alone and with the combined model, including also the engine part. The significance of the developed models is discussed, and the path towards a digital twin experiment environment is outlined. As a conclusion, we can claim that the combined simulation model is successfully constructed and shown to operate in a stable and physically plausible manner. The digital twin concept can be tested completely only when the research laboratory is constructed and ready and the test runs begin to produce measurement data for the digital part. Then also the simulation models can be tuned to a better accuracy level, and the operation as a digital twin will be verified.

Sensors doi: 10.3390/s26123853

Authors: Wei Jin Qingquan Liu Wei Dai Xin Hong Xilong Cao Haiwen Sun

Air temperature measurements in atmospheric environmental monitoring are susceptible to radiation-induced bias under natural ventilation. This study develops a low-power naturally ventilated air temperature sensor and a correction method combining computational fluid dynamics (CFD) with machine learning. The sensor integrates a Pt100 thin-film platinum resistance probe (Heraeus Holding GmbH, Hanau, Germany), symmetric guide plates, and a dual aluminum-plate radiation shield to reduce radiative heating while improving airflow around the probe. A three-dimensional fluid–solid coupled heat-transfer model was established in ANSYS FLUENT 15.0 to optimize guide-plate spacing and inclination angle and quantify the effects of solar radiation, long-wave radiation, scattered radiation, air density, wind speed, solar elevation angle, and surface albedo on radiation error. CFD results identified a guide-plate spacing of 24 mm and an inclination angle of 45° as the preferred parameters. A multilayer perceptron (MLP) model trained with CFD-derived data was validated in field experiments using a Model 076B aspirated radiation shield (Met One Instruments, Inc., Grants Pass, OR, USA) as the reference. The model predicted radiation error with a root mean square error (RMSE) of 0.052 °C, a mean absolute error (MAE) of 0.042 °C, and a correlation coefficient of 0.92. The proposed sensor and correction method provide a low-power and easy-to-maintain approach for reducing radiation-induced bias in naturally ventilated air-temperature measurements, with potential applications in meteorological observation, air-quality monitoring, and agricultural microclimate assessment.

Sensors doi: 10.3390/s26123849

Authors: Shanmugaraj Muthupandian Devendran Manoj Kumar

Wireless Body Area Networks (WBANs) have become an essential component of modern Internet of Medical Things (IoMT) healthcare systems, enabling continuous monitoring of patient physiological signals through wearable sensors. Despite their advantages, WBAN environments remain highly prone to cyber threats, privacy breaches, and single points of failure. To address these risks, this work proposes a Hybrid Multi-Metric Anomaly Detection (HM-MAD) framework deployed on the NodeMCU-32S platform with BLE 5.0 connectivity for secure continuous glucose monitoring (CGM) data transmission. The detection model simultaneously analyses physiological signals, system-level parameters, and network-level communication metrics, enabling the reliable identification of multiple cyberattacks. The proposed system focuses on securing data transmission against relay attacks, where attackers induce communication delay without modifying payloads, potentially leading to false glucose readings, improper insulin dosage delivery, unauthorized control or denial-of-service. The Convolutional Neural Network (CNN) and Bi-Directional Long Short Term Memory (BiLSTM) model classifies attack types including timing manipulation, replay attacks, power glitches, firmware tampering, and sensor spoofing. Experimental evaluation demonstrates that the proposed CNN + BiLSTM framework achieves 94.6% detection accuracy with an average inference latency of 15 ms, representing a 50% latency reduction compared to Transformer-based intrusion detection models (30 ms), while simultaneously reducing computational overhead by 28% in terms of floating-point operations and memory utilization. These results indicate that the HM-MAD framework provides an effective and scalable solution for protecting resource-constrained IoMT healthcare systems against emerging cyber threats.

Sensors doi: 10.3390/s26123852

Authors: Yaoyu Zhang Yi Xia

Single-lead electrocardiogram (ECG) is widely used in wearable devices for atrial fibrillation (AF) screening. Nevertheless, subtle pathological characteristics like P-waves and f-waves in practical signals are vulnerable to noise contamination. Meanwhile, the scarcity of high-quality annotated abnormal data instances leads to severe class imbalance. To mitigate these issues, we present an end-to-end framework designed for arrhythmia diagnosis using single-lead ECG signals, which integrates quality-aware data augmentation with a Peak-Enhanced attention mechanism. First, to mitigate the problem of data imbalance, a Quality-Aware Generative Adversarial Network (QA-GAN) is designed. This network integrates a signal quality evaluation module based on signal kurtosis, together with a dynamic soft-label training scheme, guiding the generator to prioritize learning high-quality morphological features, thereby synthesizing high-fidelity minority class samples. Second, to accurately capture subtle pathological features in electrocardiograms, a Peak-Enhanced Attention Convolutional Network (PEAC-Net) classification model is proposed. This model incorporates a Peak-Enhanced Attention (PE-Att) module, which employs learnable derivative convolutional kernels to precisely identify the transition points in the ECG signal. Furthermore, by integrating one-dimensional multi-scale dilated convolution (DSGC1D) with bidirectional LSTM, the model achieves effective capturing of both fine-grained local morphological features and long-range global rhythm patterns. Experimental results on the PhysioNet 2017 dataset indicate that the presented model attains an accuracy of 0.902 and a macro-F1 score of 0.880, respectively, outperforming other state-of-the-art models and also exhibiting robust data adaptability on the MIT-BIH dataset.

Sensors doi: 10.3390/s26123851

Authors: Kexin Guo Jingwen Wang Jiayu Lin Ningjing Chen Hengyuan Chen Zilang Zhou Manzhou Li

To address the challenges of strong signal noise, pronounced cross-modal asynchrony, high subjectivity in manually defined state labels, and insufficient model stability under extreme abnormal conditions in multi-source sensor systems, a low-noise system state prediction and anomaly detection method based on multimodal sensor signals and self-supervised representation learning is proposed. Environmental sensing data, device status data, network transmission data, operational behavior data, and event log data are uniformly modeled as system state perception signals. A temporal masking-based state structure modeling method, a state-oriented contrastive learning representation constraint mechanism, and a state representation and downstream prediction task alignment strategy are designed to learn stable, transferable, and interpretable system state features. Experimental results demonstrate that the proposed method achieves the best performance in multimodal sensor state prediction and anomaly detection tasks, with mean squared error (MSE), mean absolute error (MAE), and root mean square error (RMSE) values of 0.0167, 0.0856, and 0.1291, respectively, outperforming baseline models such as GARCH, MLP, LSTM, TCN, and Transformer. Meanwhile, IC, RankIC, and AUC reach 0.494, 0.460, and 0.815, respectively, indicating stronger state-ranking capability and improved discrimination between high-abnormality and low-abnormality states. At the classification recognition level, superior accuracy, precision, recall, and F1-score are also achieved by the proposed method, suggesting that potential abnormal states can be identified more accurately. Ablation experiments verify the effectiveness of multimodal fusion, temporal masking modeling, self-supervised contrastive constraints, and task alignment strategies. Robustness experiments further show that lower prediction errors and higher AUC can still be maintained under high-fluctuation and extreme-shock states, demonstrating strong noise resistance, stability, and practical application potential in complex sensor system scenarios.

Sensors doi: 10.3390/s26123850

Authors: Dongjing Zhang Xiaojian Hao Rui Jia Xinying Yu Yifei Fu Nengqiang Ma Zheming Cui

Sweep-field operation in a single-beam spin-exchange relaxation-free (SERF) magnetometer requires stable extraction of the dispersion zero-crossing. A frequency mismatch between the modulation signal and the demodulation references, or an unsuitable low-pass filter, can shift this zero-crossing and affect working-point determination. This paper presents a zero-crossing-stability-oriented FPGA quadrature lock-in module for SERF sweep-field demodulation. The module is designed around two requirements of sweep-field operation: maintaining a common frequency basis between the modulation output and the demodulation references, and preserving the dispersion zero-crossing when the low-pass-filter cutoff frequency is adjusted. A shared direct digital synthesizer generates both the sinusoidal modulation output and the I/Q references, keeping the excitation and demodulation signals on the same frequency basis. After quadrature multiplication, CIC decimation and a reloadable Kaiser-window FIR filter are used for low-pass processing. Board-level tests show a 1000.054 Hz spectral peak for a 1000 Hz setting and a loopback amplitude of 0.496 V, close to the ideal 0.500 V baseband amplitude. On the SERF platform, I/Q rotation reduces the quadrature residual ratio from 32.1% to 0.10%. When the FIR cutoff frequency is changed from 3 to 15 Hz, the maximum zero-crossing difference is about 0.58 ms, corresponding to 0.12% of the 2 Hz sweep period. These results show that the module supports stable zero-crossing extraction and working-point determination during sweep-field operation in a single-beam SERF magnetometer.

Sensors doi: 10.3390/s26123848

Authors: Tianfa Deng Jinchao Sun Qingjuan Zhao Faguo Huang

To address the challenges of detecting sugar orange fruits in complex natural orchard environments—where fruits are frequently occluded by leaves and branches and may be mutually occluded due to dense growth, leading to missed detections, false positives, and low detection confidence—we propose an improved algorithm based on YOLOv8n, named YOLO-FSEP. A Spatial-Channel Synergistic Attention (SCSA) module is introduced into the main network to enhance feature extraction capabilities; the IoU loss function is replaced with Focal_SIOU to improve the detection accuracy for difficult samples; and an SE attention mechanism is embedded in the detection head, with the addition of a P6 high-resolution detection layer to optimize multi-scale object performance. Experimental results on a self-built sugar orange dataset show that, compared to the baseline YOLOv8n, the improved model achieves a 0.9% increase in accuracy, a 1.3% increase in recall, and a 3.2% increase in mAP50-95, while maintaining an inference speed of 62.6 FPS. To evaluate the model under dynamic conditions, we performed a 200-frame continuous test of the 3D localization pipeline on a laptop with a RealSense D435i camera. The average YOLO inference time was 49.90 ms, post-processing (depth extraction and 3D coordinate conversion) took 0.24 ms, and the total processing time was 50.15 ms. Given that the typical response time for a robotic arm’s single positioning operation is 100–200 ms, this real-time performance meets the dynamic localization requirements of sugar orange harvesting.

Sensors doi: 10.3390/s26123846

Authors: Riccardo Gaetano Cirrone Amedeo Boldrini Alessio Polvani Xinyu Liu Francesco Vesprini Luisa Galgani Anna Witter Óscar González Gabriella Tamasi Steven Arthur Loiselle

Low-cost optical sensors have emerged as promising tools for in situ freshwater quality monitoring, offering the potential to expand spatial and temporal data coverage, particularly in community-based monitoring projects. However, despite rapid technological development of low-cost optical sensors, analytical validation practices of these devices remain poorly studied. This study aims to systematically and critically assess analytical validation practices applied to low-cost optical sensors based on absorbance, fluorescence, colorimetry, and light scattering, potentially designed for community-based freshwater monitoring. A total of 40 studies were analysed to evaluate how key analytical performance parameters, including sensitivity, accuracy, precision, and repeatability, as well as comparison with reference methods or benchtop instruments, were assessed and reported in relation to established validation guidelines. The analysis revealed substantial heterogeneity and critical gaps in validation approaches. While most studies report sensitivity metrics such as limits of detection and quantification, comprehensive evaluation of key analytical parameters such as accuracy, precision, and reproducibility was often limited. The reliance on single calibration experiments and high determination coefficients (R2) frequently overestimates sensor performance. The lack of open-source materials further limits reproducibility and deployment: essential information such as design files, calibration procedures, and open-source resources is often incomplete or unavailable. To address these limitations, we propose a structured framework for validation and reporting that integrates established analytical guidelines with the practicalities of low-cost sensor development. Adoption of this approach would enable more consistent performance evaluation, improving reproducibility and facilitating comparison across studies and devices. Overall, strengthening analytical validation and reporting practices is essential to support the transition of low-cost optical sensors from proof-of-concept systems to reliable analytical devices for freshwater quality monitoring.

Sensors doi: 10.3390/s26123847

Authors: Hairong Gu Jiaxi Wang Chenggang Chen Wenjuan Yang Mostak Ahamed Zujie Zou

Large-diameter rescue shafts serve as critical infrastructure for emergency response in mining disaster scenarios, and their structural deformation directly affects the safe passage of rescue capsules. In this paper, we investigate three-dimensional (3D) reconstruction techniques for large-diameter rescue shaft environments and develop a Neural Radiance Fields (NeRF)-based reconstruction and deformation assessment scheme. The proposed workflow integrates no reference signal-to-noise-ratio (NR-SNR), image-quality filtering, SfM-based camera-pose estimation, Nerfacto reconstruction, point-cloud export, and circular-section fitting. The NR-SNR retention-ratio experiment shows that retaining approximately 35% high-quality images provides a practical efficiency–quality trade-off for the present dataset, reducing the computational burden of SfM pose estimation while preserving sufficient geometric information for subsequent reconstruction. The reconstructed radiance field is further exported as a dense point cloud and evaluated using relative radius error, circle-fitting residuals, and image-level rendering metrics. Experiments on a simulated large-diameter rescue shaft platform show that the proposed NeRF-based scheme provides favorable geometric measurement applicability and visual reconstruction quality under weak-texture and low-illumination conditions. Compared with conventional MVS and the tested 3DGS baseline, the proposed scheme produces a point-cloud output that is more suitable for subsequent circular-section fitting and deformation-related assessment. In addition, comparison with a representative SDF-based baseline indicates that direct implicit surface recovery remains challenging for the tested hollow cylindrical shaft-wall scene. The results demonstrate the potential of the proposed NeRF-based workflow for rescue-shaft inner-wall reconstruction and engineering-oriented deformation evaluation.

Sensors doi: 10.3390/s26123845

Authors: Hyungtai Kim Mun-Taek Choi

Long-horizon mobile manipulation requires a robot to execute a sequence of heterogeneous subtasks such as navigation, picking, and articulated-object manipulation in indoor environments. Standard reinforcement learning suffers from reward sparsity and inefficient exploration in this setting, and hierarchical methods often fail at the hand-off between consecutive subtasks when the terminal state of one subtask is kinematically infeasible for the next. We propose a motion planning-augmented hierarchical reinforcement learning architecture to resolve the fundamental trade-offs between sample efficiency and hand-off reliability in long-horizon mobile manipulation. The mission is decomposed into subtasks via a Semi-Markov Decision Process; within each subtask, a collision-free reference trajectory generated by RRT* in the full joint configuration space is embedded into the reward as a per-step shaping signal; and a region-goal mechanism, defined analytically from inverse kinematics feasibility, replaces rigid coordinate hand-offs with a continuous feasible region. The architecture is evaluated in the ManiSkill-HAB simulation under teleport-free sequential execution and challenging initialization. The proposed method improves subtask success rate and sample efficiency over the baseline across all six evaluated subtasks, and the advantage compounds along the long-horizon task chain.

Sensors doi: 10.3390/s26123844

Authors: Yinghui Zhang Mao Li Zheng Lu Zitao Cai Jingzhe Shan

This work studies trustworthy use of large language models for remote sensing satellite downlink scheduling. Rather than accepting a generated optimization model at face value, we organize the workflow into three guarded steps: candidate generation, benchmark-based validation, and fallback exact solving. The core technical component is a global time-slicing validator that converts visibility windows into atomic intervals; so, mutual exclusion at the ground-station side, mutual exclusion at the satellite side, and per-satellite download caps can be checked in a physically faithful manner. Results on a prototype instance indicate that LLM-based modeling can be integrated into a dependable scheduling pipeline when external verification and recovery are built into the loop.

Sensors doi: 10.3390/s26123843

Authors: Shiquan Ling Xingchen Qin Wenkang Xu Mingmin Fu Hao Huang Shijie Ma Zhenyu Liu

As autonomous driving technology expands from structured urban roads to unstructured outdoor environments, precise understanding of complex terrain has become a critical requirement for ensuring safe vehicle navigation. However, outdoor environments are characterized by high dynamics, drastic illumination variations, ambiguous category boundaries, and prohibitive annotation costs, making traditional supervised learning methods that rely on large amounts of pixel-level annotations difficult to generalize. In this paper, we propose a novel dual-path bidirectional interactive encoder, termed LapDINO, that effectively combines the strong semantic generalization capability of the self-supervised foundation model DINOv3 with the multi-scale frequency analysis capacity of the Laplacian pyramid. Specifically, we leverage DINOv3 to extract global semantic features as a “semantic map”, while simultaneously obtaining multi-scale high-frequency details through Laplacian pyramid decomposition as “structural contours”. Building upon this, we design a bidirectional cross-attention fusion mechanism that enables dynamic interaction and mutual refinement between semantic information and geometric details. Furthermore, we introduce a multi-branch attention enhancement module that extracts pyramid features from three complementary perspectives. To address domain shift, we design lightweight visual adapters that enable efficient fine-tuning of the frozen DINOv3 backbone. Finally, we construct two off-road terrain segmentation datasets, VOTD and VOCD, to facilitate research in this domain. Experimental results demonstrate that the proposed method achieves state-of-the-art performance, striking an optimal balance between accuracy and computational efficiency, thereby providing a robust and efficient engineering solution for terrain perception in off-road environments.

Sensors doi: 10.3390/s26123842

Authors: Agnese Piersanti Gaia Maria Manes Libera Lucia Del Giudice Laura Burattini Christian Göbl Andrea Tura Micaela Morettini

Nocturnal hypoglycemia (NH) following exercise represents a critical challenge in the management of type 1 diabetes (T1D), particularly in pediatric populations, where its occurrence is associated with severe adverse outcomes and increased caregiver burden. This study aimed to identify an interpretable early signature based on CGM-derived digital biomarkers of post-exercise NH risk in children and adolescents with T1D. CGM data from 49 pediatric subjects (DirecNet cohort) were used to extract several CGM metrics across two temporal configurations: (i) Exercise + Cumulative, where features were computed over the exercise window and over an extended window spanning from exercise onset through recovery (16:00–17:00 and 16:00–22:00); and (ii) Exercise + Post-exercise, where features were computed separately over two non-overlapping intervals, capturing the exercise phase and the subsequent recovery phase (16:00–17:00 and 17:00–22:00). A Random Forest classifier was trained within a Leave-One-Out Cross Validation framework, incorporating variance inflation factor (VIF)-based multicollinearity filtering, minimum redundancy–maximum relevance (mRMR) feature selection, and SMOTE-based class balancing. The Exercise + Post-exercise configuration achieved superior performance: balanced accuracy (BA) = 76.9%, F1-score = 0.71, Area Under Receiver Operating Characteristic Curve (ROC-AUC) = 0.75, outperforming the Exercise + Cumulative configuration; this result was achieved using only five features: CONGA-15_EX (short-term glucose variability during exercise) emerged as the most robust predictor, alongside below_54 and above_250 (time spent in hypoglycemic and hyperglycemic ranges), MAG (mean absolute glucose change), and GRADE_hypo (hypoglycemia risk score). The generalizability of the temporal framework was further supported by independent validation on the OhioT1DM free-living cohort, where the Exercise + Post-exercise configuration (BA = 76.3%, ROC-AUC = 0.804) again outperformed the cumulative approach. These results suggest that a small set of interpretable CGM-derived features, extracted from the exercise and recovery windows, can effectively discriminate pediatric T1D subjects at risk of NH, supporting the development of lightweight CGM-only decision support tools for safer exercise management.

Sensors doi: 10.3390/s26123841

Authors: Younes Siraj Youssef Khardioui Youssef Mejdoub Hela Elmannai Jaouad Foshi Mohammed El Ghzaoui

This paper presents a compact wearable patch antenna operating in the 2.4 GHz ISM band for biomedical Internet of Things (IoT)-based healthcare monitoring applications. The proposed antenna is intended for integration with wearable biomedical sensors in order to support real-time physiological data transmission in remote patient monitoring systems. The antenna was designed on an FR4 substrate to achieve good impedance matching and stable radiation performance. The antenna showed good performance, with a reflection coefficient of −39.56 dB and a gain of 3.01 dB. SAR analysis confirmed compliance with IEEE and ICNIRP safety standards for wearable applications. In addition, the antenna prototype was fabricated and experimentally validated using a vector network analyzer (VNA), showing good agreement between simulated and measured results. Furthermore, the proposed system was implemented by integrating an ESP32 microcontroller with a MAX30100 physiological sensor, where the sensor is responsible for acquiring real-time biomedical data, including heart rate and blood oxygen saturation (SpO2). The ESP32 processes the acquired data and enables wireless transmission through the proposed antenna to a smartphone and laptop using the Blynk IoT platform, which allows real-time remote monitoring and visualization of physiological parameters. The obtained results confirm the suitability of the proposed antenna for wearable biomedical devices, remote healthcare monitoring, and IoT-enabled healthcare applications.

Sensors doi: 10.3390/s26123840

Authors: Hao Zhang Bingnan Wang Jiao Tong Yanan Shen

Objective: This study aimed to compare the neuromuscular control characteristics of international- and national-level squash players during forehand strokes using a multichannel surface electromyography (sEMG)-based sensing framework. By integrating wearable biosignal acquisition with muscle synergy and intermuscular coherence analyses, this study sought to identify sensor-derived markers of performance-related neuromuscular control and to provide evidence for sensor-informed squash training and athlete monitoring. Methods: Participants performed standardized forehand strokes, during which multichannel sEMG signals were synchronously collected from major upper-limb, lower-limb, and trunk muscles. The recorded sensor signals were preprocessed and analyzed using non-negative matrix factorization to extract muscle synergies, including the number of synergies, muscle weightings, and synergy activation durations. In addition, time–frequency intermuscular coherence analysis was performed on the sEMG sensor data to quantify coherence differences in the α, β, and γ frequency bands between upper-limb–trunk and lower-limb–trunk muscle pairs. Results: No significant difference was found between the two groups in the number of muscle synergies, with both groups clustering into four synergy modules. However, the sEMG sensor-based analysis revealed clear between-group differences in synergy structure and coordination patterns. International-level players showed higher muscle weightings in major proximal muscles, including the deltoid, pectoralis major, erector spinae, and gluteus maximus, and lower weightings in relatively smaller or more distal muscles such as the biceps brachii and lateral gastrocnemius. In terms of synergy timing, international-level players exhibited significantly shorter activation durations in SYN1 and SYN2, but a significantly longer activation duration in SYN3, than national-level players. For intermuscular coherence, international-level players showed significantly lower coherence in the α, β, and γ bands for multiple upper-limb–trunk and lower-limb–trunk muscle pairs. Conclusions: A multichannel sEMG sensing approach was effective in detecting performance-level differences in neuromuscular control during the squash forehand stroke. International-level players exhibited more efficient and refined neuromuscular coordination, characterized by optimized proximal muscle recruitment, more task-specific synergy timing, and reduced intermuscular coherence across selected muscle pairs. These findings highlight the value of wearable EMG sensors and sensor-based neuromuscular feature extraction for quantitative athlete assessment, movement monitoring, and the development of sensor-guided training strategies in squash.

Sensors doi: 10.3390/s26123839

Authors: Chinmaya Mutalik Rachel Redmann Sarah Bose Bryan Tassin Amy Phou Chad J. Roy

Breath aerosol analysis requires low-cost sensing substrates capable of capturing aerosolized biomolecular components while preserving chemically interpretable readouts. Here, methylcellulose–phenol red (MCPR) films are evaluated as multimodal sensing substrates using model bioaerosols consisting of sodium sulfate, bovine serum albumin (BSA), and polystyrene latex particles under acidic, neutral, and alkaline pH conditions. ATR-FTIR spectroscopy revealed inverse pH-dependent trends in sulfate (1000–1100 cm−1) and protein amide (1500–1700 cm−1) spectral regions. A sulfate-to-protein AUC ratio increased from 0.86 ± 0.01 at pH 4 to 3.56 ± 0.32 at pH 10, demonstrating ratiometric compositional discrimination of ionic and proteinaceous aerosol fractions. UV–Vis spectroscopy showed pH-dependent λmax shifts from 432 to 556 nm, confirming the preservation of phenol red optical responsiveness after aerosol exposure. FTIR-derived ratio metrics correlated linearly with optical responses, indicating coupled vibrational and optical sensing behavior. SEM-EDS analysis of methylcellulose capture films confirmed deposition of sulfate, proteinaceous, and particulate aerosol components, supporting the platform’s suitability for multimodal spectroscopic sensing. These findings establish MCPR films as integrated capture-and-sensing substrates capable of coupling optical pH responsiveness with label-free vibrational analysis, supporting future development of low-cost breath-relevant aerosol sensing platforms.

Sensors doi: 10.3390/s26123838

Authors: Manal Altaweel Judit Bisbal-Amat Juan Campos Ángel Lizana Irene Estévez

Polarimetric color cameras are a forefront technology that simultaneously captures polarimetric and color information by analyzing polarization states across different color channels, commonly red, green, and blue. In general, each of these color channels can carry different polarization information. Therefore, measuring the polarization Stokes vector at several discrete wavelengths simultaneously and with the highest possible resolution is of interest in multiple research areas. However, when a commercial color polarization sensor is used under simultaneous narrowband RGB illumination mode, its channels cannot be assumed to represent independent wavelength channels. Spectral overlap of the color filters introduces color crosstalk between wavelength-dependent analyzer intensities, which may bias the reconstructed Stokes parameters if it is not corrected before polarimetric inversion. Several methods have been proposed in the literature to address the color crosstalk problem but they typically assume that the polarization state is identical for all wavelengths. This assumption does not generally hold for real samples, which exhibit wavelength-dependent depolarization, retardance, and dichroism. To the best of our knowledge, this is the first work presenting a method that addresses the color crosstalk problem without assuming that the polarization state is identical across all wavelengths. In addition, Fourier domain demosaicking techniques are applied to interpolate the data and reconstruct the images. The present study demonstrates how the proposed method leads to an accurate recovery of chromatic and polarimetric information on both synthetic and real-world datasets. To test our approach, narrowband light beams at three wavelengths (470, 554, 630 nm), with different spatial polarization and degree of linear polarization distributions, have been simulated and validated with simulated and experimental data. The results demonstrate the feasibility of the method for accurate three polarization channels measurements.

Sensors doi: 10.3390/s26123837

Authors: Xiaojun Yin Xuyang Du Xiaohan Wu Xinming Zhang

Modern connected and automated vehicles are equipped with various onboard sensors, which continuously generate high-rate perception data. The reliable and timely sharing of such sensor data among neighboring vehicles requires high-capacity and low-latency vehicle-to-vehicle (V2V) communications. Millimeter-wave (mmWave) technology is a promising solution for supporting such high-rate transmission. However, mmWave V2V communication may be severely affected by non-line-of-sight (NLOS) blockage caused by limited transmission range, roadside obstacles, and moving vehicles. Relay forwarding can improve communication reliability and extend transmission distance, while intelligent reflecting surfaces (IRSs) can construct virtual line-of-sight (LOS) links to mitigate NLOS blockage. In this paper, we propose deploying IRSs on urban roadsides to improve mmWave multi-hop V2V communication for vehicular sensor-data sharing by integrating IRS-assisted link selection into multi-hop relay forwarding. However, IRS deployment introduces new challenges in relay selection and directional transmission coordination under interference. To address these challenges, we propose an IRS allocation and relay selection (IARS) scheme for IRS-assisted multi-hop V2V communication. The proposed scheme is based on a transmission evaluation function that jointly considers inter-vehicle distance, link quality, and concurrent transmissions. Simulation results show that the proposed IARS scheme can effectively improve communication reliability and reduce multi-hop delay, thereby supporting reliable and timely sensor-data sharing in urban vehicular networks.

Sensors doi: 10.3390/s26123836

Authors: Chenhui Xia Yeqin Shao Meiqin Che Guoqing Yang

Accurate traffic sign detection is important for the safety of autonomous driving systems. However, fully supervised methods require a large amount of manual annotation, which is cost-prohibitive and time-consuming. Semi-supervised methods employ a small amount of labeled data and a large amount of unlabeled data to train the models, hence largely reducing the annotation costs. However, these methods have the following challenges: (1) with an imbalanced long-tail class distribution of traffic signs, they tend to achieve poor performance on tail classes; (2) they often fail to detect small traffic signs. To solve these issues, we propose a Semi-Supervised Traffic Sign Detection method with Dynamic Pseudo-Label Selection and Gated Feature Fusion-based Proposal Refinement. Firstly, we design a Class Distribution-based Dynamic Pseudo-Label Selection module (CD-DPLS) to select pseudo-labels for different classes based on the class distribution information, which reduces the tendency to select more pseudo-labels from head classes instead of tail classes, thereby improving the tail class detection performance. Secondly, we employ a Gated Feature Fusion-based Proposal Refinement strategy (GFF-PR) to refine detection proposals by fusing different-scale features with a gating mechanism, which facilitates the detection of small traffic signs. In addition, we use an Adaptive-Weight Focal Loss (AWFL), with which the weight of each pseudo-label is determined by the ratio between its classification confidence and the corresponding class-specific classification-confidence threshold. Experiments on traffic sign datasets demonstrate that the proposed method outperforms state-of-the-art semi-supervised approaches, with mAP50 scores of 10.8% and 34.9% using only 1% and 10% labeled data, respectively.

Sensors doi: 10.3390/s26123835

Authors: Zhaohui Deng Zhiwu Liu Da Liu Rongjin Zhuo Xiao Yang Rong Liu

In the existing tool wear status monitoring process, the difference in the distribution of tool wear signal characteristics under different processing conditions leads to insufficient generalization of the model and poor accuracy of wear status recognition. Aiming at the problem, a method for monitoring the wear status of milling cutters under variable working conditions based on transfer learning and a lightweight SqueezeNet model is proposed. Firstly, the continuous wavelet transform (CWT) is employed to realize the conversion of the raw vibration signal to a time–frequency energy diagram to completely preserve the joint feature distribution of the vibration signal in the time and frequency dimensions. Secondly, based on the phased transfer learning strategy and the lightweight SqueezeNet, a monitoring model of the wear status of the milling cutter under variable working conditions is established, which realizes the adaptive and accurate identification of the wear status of the milling cutter under different milling conditions. Finally, comparative experiments were performed using three groups of vibration signals under different milling condition as the model inputs. As demonstrated by the experimental results, the recognition accuracy of the test set of the proposed monitoring model under variable working conditions can reach 94.583%, which is higher than the 91.133% of the LSTM-DBO-SVM model, which proves the accuracy and feasibility of the presented approach under variable working conditions.

Sensors doi: 10.3390/s26123834

Authors: Muhammad Hanif Tunio Shaowen Li Awais Ahmed Liu Lei Changyong Liang

Plant diseases result in the estimated loss of 20–40% of the world’s crop production annually, amounting to more than $220 billion in economic losses and threatening food security for a rapidly expanding world population. While the conventional methods for detecting plant diseases rely on visual inspection of the symptoms, they are resource-consuming. For effective plant disease detection at a pre-mature stage, hyperspectral imaging (HSI) represents a paradigm shift in technology. It can be used to obtain subtle spectral signatures outside the visible spectrum, which enables pre-symptomatic and highly specific plant disease diagnosis. Concurrently, deep learning (DL) has become the prevalent analytical paradigm for decoding the complex and high-dimensional data that HSI produces. This paper covers a comprehensive narrative review of the intersection of these two transformative technologies from 2008 to 2026. We first set out the biological and physical principles by which HSI is uniquely suited to detecting plant–pathogen interactions in the absence of visible symptoms. We then present a detailed taxonomy of deep learning architectures for Vision Imaging and HSI data, ranging from basic 1D and 3D convolutional neural networks (CNNs) to hybrid models with attention mechanisms and, most recently, vision transformers, which have achieved greater robustness to real-world conditions. There is currently a major and consistent “lab-to-field” performance gap. A critical analysis of various studies reveals a persistent and significant performance gap between models that perform well on controlled lab datasets (ranging from 95 to 99%) and field-collected data (typically 70–85%). This paper also addresses the practical gap of environmental variability, image noise, and the domain gap between the controlled environment and the real dataset. Finally, this review concludes by providing strategic research recommendations and a roadmap, highlighting that the future of the field is contingent upon not only architectural innovation but also a holistic approach, with robustness, scalability, affordability, and interpretability as the main focus to bring the proven potential of HSI-DL systems from the lab to the field, ultimately contributing to global food security.