Sensors

24 pages, 2584 KiB

Open AccessArticle

Precise and Continuous Biomass Measurement for Plant Growth Using a Low-Cost Sensor Setup

by Lukas Munser, Kiran Kumar Sathyanarayanan, Jonathan Raecke, Mohamed Mokhtar Mansour, Morgan Emily Uland and Stefan Streif

Sensors 2025, 25(15), 4770; https://doi.org/10.3390/s25154770 (registering DOI) - 2 Aug 2025

Abstract

Continuous and accurate biomass measurement is a critical enabler for control, decision making, and optimization in modern plant production systems. It supports the development of plant growth models for advanced control strategies like model predictive control, and enables responsive, data-driven, and plant state-dependent [...] Read more.

Continuous and accurate biomass measurement is a critical enabler for control, decision making, and optimization in modern plant production systems. It supports the development of plant growth models for advanced control strategies like model predictive control, and enables responsive, data-driven, and plant state-dependent cultivation. Traditional biomass measurement methods, such as destructive sampling, are time-consuming and unsuitable for high-frequency monitoring. In contrast, image-based estimation using computer vision and deep learning requires frequent retraining and is sensitive to changes in lighting or plant morphology. This work introduces a low-cost, load-cell-based biomass monitoring system tailored for vertical farming applications. The system operates at the level of individual growing trays, offering a valuable middle ground between impractical plant-level sensing and overly coarse rack-level measurements. Tray-level data allow localized control actions, such as adjusting light spectrum and intensity per tray, thereby enhancing the utility of controllable LED systems. This granularity supports layer-specific optimization and anomaly detection, which are not feasible with rack-level feedback. The biomass sensor is easily scalable and can be retrofitted, addressing common challenges such as mechanical noise and thermal drift. It offers a practical and robust solution for biomass monitoring in dynamic, growing environments, enabling finer control and smarter decision making in both commercial and research-oriented vertical farming systems. The developed sensor was tested and validated against manual harvest data, demonstrating high agreement with actual plant biomass and confirming its suitability for integration into vertical farming systems. Full article

(This article belongs to the Special Issue Feature Papers in Smart Agriculture 2025)

► Show Figures

Figure 1

21 pages, 875 KiB

Open AccessArticle

Comprehensive Analysis of Neural Network Inference on Embedded Systems: Response Time, Calibration, and Model Optimisation

by Patrick Huber, Ulrich Göhner, Mario Trapp, Jonathan Zender and Rabea Lichtenberg

Sensors 2025, 25(15), 4769; https://doi.org/10.3390/s25154769 (registering DOI) - 2 Aug 2025

Abstract

The response time of Artificial Neural Network (ANN) inference is critical in embedded systems processing sensor data close to the source. This is particularly important in applications such as predictive maintenance, which rely on timely state change predictions. This study enables estimation of [...] Read more.

The response time of Artificial Neural Network (ANN) inference is critical in embedded systems processing sensor data close to the source. This is particularly important in applications such as predictive maintenance, which rely on timely state change predictions. This study enables estimation of model response times based on the underlying platform, highlighting the importance of benchmarking generic ANN applications on edge devices. We analyze the impact of network parameters, activation functions, and single- versus multi-threading on response times. Additionally, potential hardware-related influences, such as clock rate variances, are discussed. The results underline the complexity of task partitioning and scheduling strategies, stressing the need for precise parameter coordination to optimise performance across platforms. This study shows that cutting-edge frameworks do not necessarily perform the required operations automatically for all configurations, which may negatively impact performance. This paper further investigates the influence of network structure on model calibration, quantified using the Expected Calibration Error (ECE), and the limits of potential optimisation opportunities. It also examines the effects of model conversion to Tensorflow Lite (TFLite), highlighting the necessity of considering both performance and calibration when deploying models on embedded systems. Full article

(This article belongs to the Section Fault Diagnosis & Sensors)

► Show Figures

Figure 1

20 pages, 12851 KiB

Open AccessArticle

Evaluation of a Vision-Guided Shared-Control Robotic Arm System with Power Wheelchair Users

by Breelyn Kane Styler, Wei Deng, Cheng-Shiu Chung and Dan Ding

Sensors 2025, 25(15), 4768; https://doi.org/10.3390/s25154768 (registering DOI) - 2 Aug 2025

Abstract

Wheelchair-mounted assistive robotic manipulators can provide reach and grasp functions for power wheelchair users. This in-lab study evaluated a vision-guided shared control (VGS) system with twelve users completing two multi-step kitchen tasks: a drinking task and a popcorn making task. Using a mixed [...] Read more.

Wheelchair-mounted assistive robotic manipulators can provide reach and grasp functions for power wheelchair users. This in-lab study evaluated a vision-guided shared control (VGS) system with twelve users completing two multi-step kitchen tasks: a drinking task and a popcorn making task. Using a mixed methods approach participants compared VGS and manual joystick control, providing performance metrics, qualitative insights, and lessons learned. Data collection included demographic questionnaires, the System Usability Scale (SUS), NASA Task Load Index (NASA-TLX), and exit interviews. No significant SUS differences were found between control modes, but NASA-TLX scores revealed VGS control significantly reduced workload during the drinking task and the popcorn task. VGS control reduced operation time and improved task success but was not universally preferred. Six participants preferred VGS, five preferred manual, and one had no preference. In addition, participants expressed interest in robotic arms for daily tasks and described two main operation challenges: distinguishing wrist orientation from rotation modes and managing depth perception. They also shared perspectives on how a personal robotic arm could complement caregiver support in their home. Full article

(This article belongs to the Special Issue Intelligent Sensors and Robots for Ambient Assisted Living)

► Show Figures

Figure 1

16 pages, 5536 KiB

Open AccessArticle

The Development of a Wearable-Based System for Detecting Shaken Baby Syndrome Using Machine Learning Models

by Ram Kinker Mishra, Khalid Al Ansari, Rylee Cole, Arin Nazarian, Ilkay Yildiz Potter and Ashkan Vaziri

Sensors 2025, 25(15), 4767; https://doi.org/10.3390/s25154767 (registering DOI) - 2 Aug 2025

Abstract

Shaken Baby Syndrome (SBS) is one of the primary causes of fatal head trauma in infants and young children, occurring in about 33 per 100,000 infants annually in the U.S., with mortality rates being between 15% and 38%. Survivors frequently endure long-term disabilities, [...] Read more.

Shaken Baby Syndrome (SBS) is one of the primary causes of fatal head trauma in infants and young children, occurring in about 33 per 100,000 infants annually in the U.S., with mortality rates being between 15% and 38%. Survivors frequently endure long-term disabilities, such as cognitive deficits, visual impairments, and motor dysfunction. Diagnosing SBS remains difficult due to the lack of visible injuries and delayed symptom onset. Existing detection methods—such as neuroimaging, biomechanical modeling, and infant monitoring systems—cannot perform real-time detection and face ethical, technical, and accuracy limitations. This study proposes an inertial measurement unit (IMU)-based detection system enhanced with machine learning to identify aggressive shaking patterns. Findings indicate that wearable-based motion analysis is a promising method for recognizing high-risk shaking, offering a non-invasive, real-time solution that could minimize infant harm and support timely intervention. Full article

(This article belongs to the Special Issue Smart Wearable Health Monitoring Systems: Materials, Sensors, Nanogenerators and Self-Powered Applications)

► Show Figures

Figure 1

14 pages, 654 KiB

Open AccessArticle

A Conceptual Framework for User Trust in AI Biosensors: Integrating Cognition, Context, and Contrast

by Andrew Prahl

Sensors 2025, 25(15), 4766; https://doi.org/10.3390/s25154766 (registering DOI) - 2 Aug 2025

Abstract

Artificial intelligence (AI) techniques have propelled biomedical sensors beyond measuring physiological markers to interpreting subjective states like stress, pain, or emotions. Despite these technological advances, user trust is not guaranteed and is inadequately addressed in extant research. This review proposes the Cognition–Context–Contrast (CCC) [...] Read more.

Artificial intelligence (AI) techniques have propelled biomedical sensors beyond measuring physiological markers to interpreting subjective states like stress, pain, or emotions. Despite these technological advances, user trust is not guaranteed and is inadequately addressed in extant research. This review proposes the Cognition–Context–Contrast (CCC) conceptual framework to explain the trust and acceptance of AI-enabled sensors. First, we map cognition, comprising the expectations and stereotypes that humans have about machines. Second, we integrate task context by situating sensor applications along an intellective-to-judgmental continuum and showing how demonstrability predicts tolerance for sensor uncertainty and/or errors. Third, we analyze contrast effects that arise when automated sensing displaces familiar human routines, heightening scrutiny and accelerating rejection if roll-out is abrupt. We then derive practical implications such as enhancing interpretability, tailoring data presentations to task demonstrability, and implementing transitional introduction phases. The framework offers researchers, engineers, and clinicians a structured conceptual framework for designing and implementing the next generation of AI biosensors. Full article

(This article belongs to the Special Issue AI in Sensor-Based E-Health, Wearables and Assisted Technologies)

► Show Figures

Figure 1

15 pages, 1612 KiB

Open AccessArticle

Flexible Strain Sensor Based on PVA/Tannic Acid/Lithium Chloride Ionically Conductive Hydrogel with Excellent Sensing and Good Adhesive Properties

by Xuanyu Pan, Hongyuan Zhu, Fufei Qin, Mingxing Jing, Han Wu and Zhuangzhi Sun

Sensors 2025, 25(15), 4765; https://doi.org/10.3390/s25154765 (registering DOI) - 1 Aug 2025

Abstract

Ion-conductive-hydrogel strain sensors demonstrate broad application prospects in the fields of flexible sensing and bioelectric signal monitoring due to their excellent skin conformability and efficient signal transmission characteristics. However, traditional preparation methods face significant challenges in enhancing adhesion strength, conductivity, and mechanical stability. [...] Read more.

Ion-conductive-hydrogel strain sensors demonstrate broad application prospects in the fields of flexible sensing and bioelectric signal monitoring due to their excellent skin conformability and efficient signal transmission characteristics. However, traditional preparation methods face significant challenges in enhancing adhesion strength, conductivity, and mechanical stability. To address this issue, this study employed a freeze–thaw cycling method, using polyvinyl alcohol (PVA) as the matrix material, tannic acid (TA) as the adhesion reinforcement material, and lithium chloride (LiCl) as the conductive medium, successfully developing an ion-conductive hydrogel with superior comprehensive performance. Experimental data confirm that the PVA-TA-0.5/LiCl-1 hydrogel achieves optimal levels of adhesion strength (2.32 kPa on pigskin) and conductivity (0.64 S/m), while also exhibiting good tensile strength (0.1 MPa). Therefore, this hydrogel shows great potential for use in strain sensors, demonstrating excellent sensitivity (GF = 1.15), reliable operational stability, as the ΔR/R₀ signal remains virtually unchanged after 2500 cycles of stretching, and outstanding strain sensing and electromyographic signal acquisition capabilities, fully highlighting its practical value in the fields of flexible sensing and bioelectric monitoring. Full article

(This article belongs to the Section Sensor Materials)

20 pages, 5650 KiB

Open AccessArticle

The In-Plane Deformation and Free Vibration Analysis of a Rotating Ring Resonator of a Gyroscope with Evenly Distributed Mass Imperfections

by Dongsheng Zhang and Shuming Li

Sensors 2025, 25(15), 4764; https://doi.org/10.3390/s25154764 (registering DOI) - 1 Aug 2025

Abstract

A rotating imperfect ring resonator of the gyroscope is modeled by a rotating thin ring with evenly distributed point masses. The free response of the rotating ring structure at constant speed is investigated, including the steady elastic deformation and wave response. The dynamic [...] Read more.

A rotating imperfect ring resonator of the gyroscope is modeled by a rotating thin ring with evenly distributed point masses. The free response of the rotating ring structure at constant speed is investigated, including the steady elastic deformation and wave response. The dynamic equations are formulated by using Hamilton’s principle in the ground-fixed coordinates. The coordinate transformation is applied to facilitate the solution of the steady deformation, and the displacements and tangential tension for the deformation are calculated by the perturbation method. Employing Galerkin’s method, the governing equation of the free vibration is casted in matrix differential operator form after the separation of the real and imaginary parts with the inextensional assumption. The natural frequencies are calculated through the eigenvalue analysis, and the numerical results are obtained. The effects of the point masses on the natural frequencies of the forward and backward traveling wave curves of different orders are discussed, especially on the measurement accuracy of gyroscopes for different cases. In the ground-fixed coordinates, the frequency splitting results in a crosspoint of the natural frequencies of the forward and backward traveling waves. The finite element method is applied to demonstrate the validity and accuracy of the model. Full article

(This article belongs to the Section Physical Sensors)

► Show Figures

Graphical abstract

25 pages, 2860 KiB

Open AccessReview

Multimodal Sensing-Enabled Large Language Models for Automated Emotional Regulation: A Review of Current Technologies, Opportunities, and Challenges

by Liangyue Yu, Yao Ge, Shuja Ansari, Muhammad Imran and Wasim Ahmad

Sensors 2025, 25(15), 4763; https://doi.org/10.3390/s25154763 (registering DOI) - 1 Aug 2025

Abstract

Emotion regulation is essential for mental health. However, many people ignore their own emotional regulation or are deterred by the high cost of psychological counseling, which poses significant challenges to making effective support widely available. This review systematically examines the convergence of multimodal [...] Read more.

Emotion regulation is essential for mental health. However, many people ignore their own emotional regulation or are deterred by the high cost of psychological counseling, which poses significant challenges to making effective support widely available. This review systematically examines the convergence of multimodal sensing technologies and large language models (LLMs) for the development of Automated Emotional Regulation (AER) systems. The review draws upon a comprehensive analysis of the existing literature, encompassing research papers, technical reports, and relevant theoretical frameworks. Key findings indicate that multimodal sensing offers the potential for rich, contextualized data pertaining to emotional states, while LLMs provide improved capabilities for interpreting these inputs and generating nuanced, empathetic, and actionable regulatory responses. The integration of these technologies, including physiological sensors, behavioral tracking, and advanced LLM architectures, presents the improvement of application, moving AER beyond simpler, rule-based systems towards more adaptive, context-aware, and human-like interventions. Opportunities for personalized interventions, real-time support, and novel applications in mental healthcare and other domains are considerable. However, these prospects are counterbalanced by significant challenges and limitations. In summary, this review synthesizes current technological advancements, identifies substantial opportunities for innovation and application, and critically analyzes the multifaceted technical, ethical, and practical challenges inherent in this domain. It also concludes that while the integration of multimodal sensing and LLMs holds significant potential for AER, the field is nascent and requires concerted research efforts to realize its full capacity to enhance human well-being. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Figure 1

21 pages, 4517 KiB

Open AccessArticle

A Method Integrating the Matching Field Algorithm for the Three-Dimensional Positioning and Search of Underwater Wrecked Targets

by Huapeng Cao, Tingting Yang and Ka-Fai Cedric Yiu

Sensors 2025, 25(15), 4762; https://doi.org/10.3390/s25154762 (registering DOI) - 1 Aug 2025

Abstract

In this paper, a joint Matching Field Processing (MFP) Algorithm based on horizontal uniform circular array (UCA) is proposed for three-dimensional position of underwater wrecked targets. Firstly, a Marine search and rescue position model based on Minimum Variance Distortionless Response (MVDR) and matching [...] Read more.

In this paper, a joint Matching Field Processing (MFP) Algorithm based on horizontal uniform circular array (UCA) is proposed for three-dimensional position of underwater wrecked targets. Firstly, a Marine search and rescue position model based on Minimum Variance Distortionless Response (MVDR) and matching field quadratic joint Algorithm was proposed. Secondly, an MVDR beamforming method based on pre-Kalman filtering is designed to refine the real-time DOA estimation of the desired signal and the interference source, and the sound source azimuth is determined for prepositioning. The antenna array weights are dynamically adjusted according to the filtered DOA information. Finally, the Adaptive Matching Field Algorithm (AMFP) used the DOA information to calculate the range and depth of the lost target, and obtained the range and depth estimates. Thus, the 3D position of the lost underwater target is jointly estimated. This method alleviates the angle ambiguity problem and does not require a computationally intensive 2D spectral search. The simulation results show that the proposed method can better realise underwater three-dimensional positioning under certain signal-to-noise ratio conditions. When there is no error in the sensor coordinates, the positioning error is smaller than that of the baseline method as the SNR increases. When the SNR is 0 dB, with the increase in the sensor coordinate error, the target location error increases but is smaller than the error amplitude of the benchmark Algorithm. The experimental results verify the robustness of the proposed framework in the hierarchical ocean environment, which provides a practical basis for the deployment of rapid response underwater positioning systems in maritime search and rescue scenarios. Full article

(This article belongs to the Special Issue Sensor Fusion in Positioning and Navigation)

15 pages, 3678 KiB

Open AccessArticle

Virtual Signal Processing-Based Integrated Multi-User Detection

by Dabao Wang and Zhao Li

Sensors 2025, 25(15), 4761; https://doi.org/10.3390/s25154761 (registering DOI) - 1 Aug 2025

Abstract

The demand for high data rates and large system capacity has posed significant challenges for medium access control (MAC) methods. Successive interference cancellation (SIC) is a classical multi-user detection (MUD) method; however, it suffers from an error propagation problem. To address this deficiency, [...] Read more.

The demand for high data rates and large system capacity has posed significant challenges for medium access control (MAC) methods. Successive interference cancellation (SIC) is a classical multi-user detection (MUD) method; however, it suffers from an error propagation problem. To address this deficiency, we propose a method called Virtual Signal Processing-Based Integrated Multi-User Detection (VSP-IMUD). In VSP-IMUD, the received mixed multi-user signals are treated as an equivalent signal. The channel ambiguity corresponding to each user’s signal is then examined. For channels with non-zero ambiguity values, the signal components are detected using zero-forcing (ZF) reception. Next, the detected ambiguous signal components are reconstructed and subtracted from the received mixed signal using SIC. Once all the ambiguous signals are detected, the remaining signal components with zero ambiguity values are equated to a virtual integrated signal, to which a matched filter (MF) is applied. Finally, by selecting the signal with the highest channel gain and adopting its data as the reference symbol, the remaining signals’ dataset can be determined. Our theoretical analysis and simulation results demonstrate that VSP-IMUD effectively reduces the frequency of SIC applications and mitigates its error propagation effects, thereby improving the system’s bit-error rate (BER) performance. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Figure 1

22 pages, 2449 KiB

Open AccessArticle

Tracking Consensus for Nonlinear Multi-Agent Systems Under Asynchronous Switching and Undirected Topology

by Shanyan Hu and Mengling Wang

Sensors 2025, 25(15), 4760; https://doi.org/10.3390/s25154760 (registering DOI) - 1 Aug 2025

Abstract

This paper investigates the tracking consensus of nonlinear multi-agent systems under undirected topology, considering asynchronous switching caused by delays between communication topology switching and controller switching. First, by using the properties of undirected topology graphs, the controller design process is simplified. Then, to [...] Read more.

This paper investigates the tracking consensus of nonlinear multi-agent systems under undirected topology, considering asynchronous switching caused by delays between communication topology switching and controller switching. First, by using the properties of undirected topology graphs, the controller design process is simplified. Then, to address asynchronous delays during topology switching, the system operation is divided into synchronized and delayed modes based on the status of the controller and topology. Every operating mode has a corresponding control strategy. To alleviate the burden of communication and computation, an event-triggered mechanism (ETM) is introduced to reduce the number of controller updates. By constructing an augmented Lyapunov function that incorporates both matching and mismatching periods, sufficient conditions ensuring system stability are established. The required controller based on the dynamic ETM is obtained by solving Linear Matrix Inequalities (LMIs). Finally, a simulation example is conducted to verify its effectiveness. Full article

(This article belongs to the Special Issue Wireless Communication Technologies for Internet of Things and Wireless Sensor Networks)

► Show Figures

Figure 1

23 pages, 3467 KiB

Open AccessArticle

Resampling Multi-Resolution Signals Using the Bag of Functions Framework: Addressing Variable Sampling Rates in Time Series Data

by David Orlando Salazar Torres, Diyar Altinses and Andreas Schwung

Sensors 2025, 25(15), 4759; https://doi.org/10.3390/s25154759 (registering DOI) - 1 Aug 2025

Abstract

In time series analysis, the ability to effectively handle data with varying sampling rates is crucial for accurate modeling and analysis. This paper presents the MR-BoF (Multi-Resolution Bag of Functions) framework, which leverages sampling-rate-independent techniques to decompose time series data while accommodating signals [...] Read more.

In time series analysis, the ability to effectively handle data with varying sampling rates is crucial for accurate modeling and analysis. This paper presents the MR-BoF (Multi-Resolution Bag of Functions) framework, which leverages sampling-rate-independent techniques to decompose time series data while accommodating signals with differing resolutions. Unlike traditional methods that require uniform sampling frequencies, the BoF framework employs a flexible encoding approach, allowing for the integration of multi-resolution time series. Through a series of experiments, we demonstrate that the BoF framework ensures the precise reconstruction of the original data while enhancing resampling capabilities by utilizing decomposed components. The results show that this method offers significant advantages in scenarios involving irregular sampling rates and heterogeneous acquisition systems, making it a valuable tool for applications in fields such as finance, healthcare, industrial monitoring, IoT networks, and sensor networks. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Graphical abstract

18 pages, 3271 KiB

Open AccessArticle

Mobile App–Induced Mental Fatigue Affects Strength Asymmetry and Neuromuscular Performance Across Upper and Lower Limbs

by Andreas Stafylidis, Walter Staiano, Athanasios Mandroukas, Yiannis Michailidis, Lluis Raimon Salazar Bonet, Marco Romagnoli and Thomas I. Metaxas

Sensors 2025, 25(15), 4758; https://doi.org/10.3390/s25154758 (registering DOI) - 1 Aug 2025

Abstract

This study aimed to investigate the effects of mental fatigue on physical and cognitive performance (lower-limb power, isometric and handgrip strength, and psychomotor vigilance). Twenty-two physically active young adults (12 males, 10 females; M_age = 20.82 ± 1.47) were randomly assigned to [...] Read more.

This study aimed to investigate the effects of mental fatigue on physical and cognitive performance (lower-limb power, isometric and handgrip strength, and psychomotor vigilance). Twenty-two physically active young adults (12 males, 10 females; M_age = 20.82 ± 1.47) were randomly assigned to either a Mental Fatigue (MF) or Control group (CON). The MF group showed a statistically significant (p = 0.019) reduction in non-dominant handgrip strength, declining by approximately 2.3 kg (about 5%), while no such change was observed in the CON group or in dominant handgrip strength across groups. Reaction time (RT) was significantly impaired following the mental fatigue protocol: RT increased by 117.82 ms, representing an approximate 46% longer response time in the MF group (p < 0.001), whereas the CON group showed a smaller, non-significant increase of 32.82 ms (~12% longer). No significant differences were found in squat jump performance, indicating that lower-limb explosive power may be less affected by acute mental fatigue. These findings demonstrate that mental fatigue selectively impairs fine motor strength and cognitive processing speed, particularly reaction time, while gross motor power remains resilient. Understanding these effects is critical for optimizing performance in contexts requiring fine motor control and sustained attention under cognitive load. Full article

(This article belongs to the Special Issue Sensing Human Cognitive Factors)

► Show Figures

Figure 1

25 pages, 2859 KiB

Open AccessArticle

Feature-Based Normality Models for Anomaly Detection

by Hui Yie Teh, Kevin I-Kai Wang and Andreas W. Kempa-Liehr

Sensors 2025, 25(15), 4757; https://doi.org/10.3390/s25154757 (registering DOI) - 1 Aug 2025

Abstract

Detecting previously unseen anomalies in sensor data is a challenging problem for artificial intelligence when sensor-specific and deployment-specific characteristics of the time series need to be learned from a short calibration period. From the application point of view, this challenge becomes increasingly important [...] Read more.

Detecting previously unseen anomalies in sensor data is a challenging problem for artificial intelligence when sensor-specific and deployment-specific characteristics of the time series need to be learned from a short calibration period. From the application point of view, this challenge becomes increasingly important because many applications are gravitating towards utilising low-cost sensors for Internet of Things deployments. While these sensors offer cost-effectiveness and customisation, their data quality does not match that of their high-end counterparts. To improve sensor data quality while addressing the challenges of anomaly detection in Internet of Things applications, we present an anomaly detection framework that learns a normality model of sensor data. The framework models the typical behaviour of individual sensors, which is crucial for the reliable detection of sensor data anomalies, especially when dealing with sensors observing significantly different signal characteristics. Our framework learns sensor-specific normality models from a small set of anomaly-free training data while employing an unsupervised feature engineering approach to select statistically significant features. The selected features are subsequently used to train a Local Outlier Factor anomaly detection model, which adaptively determines the boundary separating normal data from anomalies. The proposed anomaly detection framework is evaluated on three real-world public environmental monitoring datasets with heterogeneous sensor readings. The sensor-specific normality models are learned from extremely short calibration periods (as short as the first 3 days or 10% of the total recorded data) and outperform four other state-of-the-art anomaly detection approaches with respect to F1-score (between 5.4% and 9.3% better) and Matthews correlation coefficient (between 4.0% and 7.6% better). Full article

(This article belongs to the Special Issue Innovative Approaches to Cybersecurity for IoT and Wireless Networks)

► Show Figures

Figure 1

14 pages, 2795 KiB

Open AccessArticle

Obtaining Rotational Stiffness of Wind Turbine Foundation from Acceleration and Wind Speed SCADA Data

by Jiazhi Dai, Mario Rotea and Nasser Kehtarnavaz

Sensors 2025, 25(15), 4756; https://doi.org/10.3390/s25154756 (registering DOI) - 1 Aug 2025

Abstract

Monitoring the health of wind turbine foundations is essential for ensuring their operational safety. This paper presents a cost-effective approach to obtain rotational stiffness of wind turbine foundations by using only acceleration and wind speed data that are part of SCADA data, thus [...] Read more.

Monitoring the health of wind turbine foundations is essential for ensuring their operational safety. This paper presents a cost-effective approach to obtain rotational stiffness of wind turbine foundations by using only acceleration and wind speed data that are part of SCADA data, thus lowering the use of moment and tilt sensors that are currently being used for obtaining foundation stiffness. First, a convolutional neural network model is applied to map acceleration and wind speed data within a moving window to corresponding moment and tilt values. Rotational stiffness of the foundation is then estimated by fitting a line in the moment-tilt plane. The results obtained indicate that such a mapping model can provide stiffness values that are within 7% of ground truth stiffness values on average. Second, the developed mapping model is re-trained by using synthetic acceleration and wind speed data that are generated by an autoencoder generative AI network. The results obtained indicate that although the exact amount of stiffness drop cannot be determined, the drops themselves can be detected. This mapping model can be used not only to lower the cost associated with obtaining foundation rotational stiffness but also to sound an alarm when a foundation starts deteriorating. Full article

(This article belongs to the Special Issue Sensors Technology Applied in Power Systems and Energy Management)

► Show Figures

Figure 1

14 pages, 21956 KiB

Open AccessArticle

Evaluating Image Quality Metrics as Loss Functions for Image Dehazing

by Rareș Dobre-Baron, Adrian Savu-Jivanov and Cosmin Ancuți

Sensors 2025, 25(15), 4755; https://doi.org/10.3390/s25154755 (registering DOI) - 1 Aug 2025

Abstract

The difficulty and manual nature of procuring human evaluators for ranking the quality of images affected by various types of degradations, and of those cleaned up by developed algorithms, has lead to the widespread adoption of automated metrics, like the Peak Signal-to-Noise Ratio [...] Read more.

The difficulty and manual nature of procuring human evaluators for ranking the quality of images affected by various types of degradations, and of those cleaned up by developed algorithms, has lead to the widespread adoption of automated metrics, like the Peak Signal-to-Noise Ratio (PSNR) and the Structural Similarity Index Metric (SSIM). However, disparities between rankings given by these metrics and those given by human evaluators have encouraged the development of improved image quality assessment (IQA) metrics that are a better fit for this purpose. These methods have been previously used solely for quality assessments and not as objectives in the training of neural networks for high-level vision tasks, despite the potential improvements that may come about by directly optimizing for desired metrics. This paper examines the adequacy of ten recent IQA metrics, compared with standard loss functions, within two trained dehazing neural networks, with observed broad improvement in their performance. Full article

(This article belongs to the Special Issue Sensing and Imaging in Computer Vision)

20 pages, 4782 KiB

Open AccessArticle

Enhanced Spatiotemporal Landslide Displacement Prediction Using Dynamic Graph-Optimized GNSS Monitoring

by Jiangfeng Li, Jiahao Qin, Kaimin Kang, Mingzhi Liang, Kunpeng Liu and Xiaohua Ding

Sensors 2025, 25(15), 4754; https://doi.org/10.3390/s25154754 (registering DOI) - 1 Aug 2025

Abstract

Landslide displacement prediction is crucial for disaster mitigation, yet traditional methods often fail to capture the complex, non-stationary spatiotemporal dynamics of slope evolution. This study introduces an enhanced prediction framework that integrates multi-scale signal processing with dynamic, geology-aware graph modeling. The proposed methodology [...] Read more.

Landslide displacement prediction is crucial for disaster mitigation, yet traditional methods often fail to capture the complex, non-stationary spatiotemporal dynamics of slope evolution. This study introduces an enhanced prediction framework that integrates multi-scale signal processing with dynamic, geology-aware graph modeling. The proposed methodology first employs the Maximum Overlap Discrete Wavelet Transform (MODWT) to denoise raw Global Navigation Satellite System (GNSS)-monitored displacement time series data, enhancing the underlying deformation features. Subsequently, a geology-aware graph is constructed, using the temporal correlation of displacement series as a practical proxy for physical relatedness between monitoring nodes. The framework’s core innovation lies in a dynamic graph optimization model with low-rank constraints, which adaptively refines the graph topology to reflect time-varying inter-sensor dependencies driven by factors like mining activities. Experiments conducted on a real-world dataset from an active open-pit mine demonstrate the framework’s superior performance. The DCRNN-proposed model achieved the highest accuracy among eight competing models, recording a Root Mean Square Error (RMSE) of 2.773 mm in the Vertical direction, a 39.1% reduction compared to its baseline. This study validates that the proposed dynamic graph optimization approach provides a robust and significantly more accurate solution for landslide prediction in complex, real-world engineering environments. Full article

(This article belongs to the Section Navigation and Positioning)

► Show Figures

Figure 1

13 pages, 1717 KiB

Open AccessArticle

High-Performance Hydrogen Gas Sensor Based on Pd-Doped MoS₂/Si Heterojunction

by Enyu Ma, Zihao Xu, Ankai Sun, Shuo Yang and Jianyu Jiang

Sensors 2025, 25(15), 4753; https://doi.org/10.3390/s25154753 (registering DOI) - 1 Aug 2025

Abstract

High-performance hydrogen gas sensors have gained considerable interest for their crucial function in reducing H₂ explosion risk. Although MoS₂ has good potential for chemical sensing, its application in hydrogen detection at room temperature is limited by slow response and incomplete recovery. [...] Read more.

High-performance hydrogen gas sensors have gained considerable interest for their crucial function in reducing H₂ explosion risk. Although MoS₂ has good potential for chemical sensing, its application in hydrogen detection at room temperature is limited by slow response and incomplete recovery. In this work, Pd-doped MoS₂ thin films are deposited on a Si substrate, forming Pd-doped MoS₂/Si heterojunctions via magnetron co-sputtering. The incorporation of Pd nanoparticles significantly enhances the catalytic activity for hydrogen adsorption and facilitates more efficient electron transfer. Owing to its distinct structural characteristics and sharp interface properties, the fabricated Pd-doped MoS₂/Si heterojunction device exhibits excellent H₂ sensing performance under room temperature conditions. The gas sensor device achieves an impressive sensing response of ~6.4 × 10³% under 10,000 ppm H₂ concentration, representing a 110% improvement compared to pristine MoS₂. Furthermore, the fabricated heterojunction device demonstrates rapid response and recovery times (24.6/12.2 s), excellent repeatability, strong humidity resistance, and a ppb-level detection limit. These results demonstrate the promising application prospects of Pd-doped MoS₂/Si heterojunctions in the development of advanced gas sensing devices. Full article

(This article belongs to the Special Issue 2D Materials for Advanced Sensing Technology)

► Show Figures

Figure 1

23 pages, 3580 KiB

Open AccessArticle

Distributed Collaborative Data Processing Framework for Unmanned Platforms Based on Federated Edge Intelligence

by Siyang Liu, Nanliang Shan, Xianqiang Bao and Xinghua Xu

Sensors 2025, 25(15), 4752; https://doi.org/10.3390/s25154752 (registering DOI) - 1 Aug 2025

Abstract

Unmanned platforms such as unmanned aerial vehicles, unmanned ground vehicles, and autonomous underwater vehicles often face challenges of data, device, and model heterogeneity when performing collaborative data processing tasks. Existing research does not simultaneously address issues from these three aspects. To address this [...] Read more.

Unmanned platforms such as unmanned aerial vehicles, unmanned ground vehicles, and autonomous underwater vehicles often face challenges of data, device, and model heterogeneity when performing collaborative data processing tasks. Existing research does not simultaneously address issues from these three aspects. To address this issue, this study designs an unmanned platform cluster architecture inspired by the cloud-edge-end model. This architecture integrates federated learning for privacy protection, leverages the advantages of distributed model training, and utilizes edge computing’s near-source data processing capabilities. Additionally, this paper proposes a federated edge intelligence method (DSIA-FEI), which comprises two key components. Based on traditional federated learning, a data sharing mechanism is introduced, in which data is extracted from edge-side platforms and placed into a data sharing platform to form a public dataset. At the beginning of model training, random sampling is conducted from the public dataset and distributed to each unmanned platform, so as to mitigate the impact of data distribution heterogeneity and class imbalance during collaborative data processing in unmanned platforms. Moreover, an intelligent model aggregation strategy based on similarity measurement and loss gradient is developed. This strategy maps heterogeneous model parameters to a unified space via hierarchical parameter alignment, and evaluates the similarity between local and global models of edge devices in real-time, along with the loss gradient, to select the optimal model for global aggregation, reducing the influence of device and model heterogeneity on cooperative learning of unmanned platform swarms. This study carried out extensive validation on multiple datasets, and the experimental results showed that the accuracy of the DSIA-FEI proposed in this paper reaches 0.91, 0.91, 0.88, and 0.87 on the FEMNIST, FEAIR, EuroSAT, and RSSCN7 datasets, respectively, which is more than 10% higher than the baseline method. In addition, the number of communication rounds is reduced by more than 40%, which is better than the existing mainstream methods, and the effectiveness of the proposed method is verified. Full article

(This article belongs to the Special Issue Signal Processing and Machine Learning Techniques for Intelligent Sensing Applications)

► Show Figures

Figure 1

Journal Description

Sensors

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI