Sensors最新文献

To alleviate communication pressure and terminal resource constraints in mobile edge computing (MEC) networks, this paper proposes a resource allocation optimization method for MEC systems that integrates data compression technology and non-orthogonal multiple access technology. This method considers practical constraints such as terminal device battery capacity and computational resource limitations. By jointly optimizing computational resource allocation, task offloading strategies, and data compression ratios, it constructs an optimization model aimed at minimizing the total task processing latency. Addressing the challenges stemming from the non-convex nature of the problem and the dynamic variations in channel conditions and task requirements, this paper proposes a softmax deep double deterministic policy gradient algorithm, where softmax operator function mitigates both overestimation and underestimation biases inherent in traditional reinforcement learning frameworks, enhancing convergence performance. Utilizing a deep reinforcement learning framework, the algorithm achieves joint decision-making optimization for computational resources, task offloading, and compression ratios, thereby minimizing the total task processing latency while satisfying transmit power and computational resource constraints. Simulation results demonstrate that the proposed scheme exhibits significant advantages over benchmark algorithms in terms of convergence speed and task processing latency.

In 2025, there is still no ubiquitous, accurate, infrastructure-free indoor positioning system. Among existing approaches, WiFi-based positioning is highly promising as it leverages existing infrastructure. However, its performance is severely affected by WiFi signal variability and environmental dynamics. Thus, this paper presents a novel approach that combines WiFi Round Trip Time and Received Signal Strength measurements with the Conformal Prediction (CP) framework to achieve robust uncertainty-aware indoor positioning. Our proposed method does not only accurately estimate the user position, but also provides two prediction regions: a rectangular region and a circular region. We systematically evaluate our method across three real-world testbeds, which achieves a positioning accuracy of 0.6 m, while generating prediction regions with theoretical coverage guarantees for circular regions and marginal coverage guarantees for rectangular regions. To the best of our knowledge, this is one of the first work to enable uncertainty quantification on top of state-of-the-art WiFi ranging signals.

The variable-pitch connecting rod of a helicopter bears axial tensile and compressive loads during operation. The traditional load monitoring method using strain gauge is easily affected by external conditions. Therefore, a giant magnetostrictive (GM) tension and compression force sensor with permanent magnet bias is proposed and optimized. Because the bias magnetic field plays a decisive role in the performance of the sensor, this paper has carried out in-depth research on this. Firstly, the mathematical model of the magnetic circuit is established, and the various magnetic circuits of the sensor are simulated and analyzed. Secondly, the magnetic flux uniformity of the GMM rod is used as the evaluation index, and the relative permeability of the magnetic material and the structure are systematically studied. The influence of parameters on the magnetic flux of the magnetic circuit, and finally the optimal parameter combination of the magnetic circuit is determined by orthogonal test. The results show that when the magnetic circuit without the magnetic side wall is used, the magnetic material can better guide the magnetic flux through the GMM rod; the magnetic flux uniformity of the optimized GMM force sensor is increased by 7.44%, the magnetic flux density is increased by 13.9 mT and the Hall output voltage increases linearly by 1.125% in the same proportion. This provides an important reference for improving the utilization rate of GMM rods and also improves the safety of flight operation and reduces maintenance costs.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive and motor decline. Early detection remains challenging, as traditional neuroimaging and neuropsychological assessments often fail to capture subtle, preclinical changes. Recent advances in digital health and artificial intelligence (AI) offer new opportunities to identify non-invasive biomarkers of cognitive impairment. In this study, we propose an AI-driven framework for early AD based on handwriting motion data captured using a sensor-integrated Smart Pen. The system employs an inertial measurement unit (MPU-9250) to record fine-grained kinematic and dynamic signals during handwriting and drawing tasks. Multiple machine learning (ML) algorithms-Logistic Regression, Support Vector Machine (SVM), Random Forest (RF), and k-Nearest Neighbors (kNN)-and deep learning (DL) architectures, including one-dimensional Convolutional Neural Networks (1D-CNN), Long Short-Term Memory (LSTM), and a hybrid CNN-BiLSTM network, were systematically evaluated. To address data scarcity, we implemented a Sim-to-Real Domain Adaptation strategy, augmenting the training set with physics-based synthetic samples. Results show that classical ML models achieved moderate diagnostic performance (AUC: 0.62-0.76), while the proposed hybrid DL model demonstrated superior predictive capability (accuracy: 0.91, AUC: 0.96). These findings underscore the potential of motion-based digital biomarkers for the automated, non-invasive detection of AD. The proposed framework represents a cost-effective and clinically scalable informatics solution for digital cognitive assessment.

The Sparse Subgraph Finding (SGF) problem addresses the challenge of identifying sub-graphs with weak social interactions and sparse connections within a graph, which can be effectively modeled as discovering sparse subsystems in intelligent sensor networks. Traditional methods often rely on manually designed heuristics, which are computationally expensive and lack scalability, especially when dealing with complex sensor network systems. In this paper, we propose RL-SGF, a novel framework that integrates deep reinforcement learning and graph embedding through joint optimization to overcome these limitations. By simultaneously optimizing subsystem sparsity and representation learning within a unified framework, RL-SGF enhances both the effectiveness and robustness of the model in sensor network applications. Experimental results on synthetic and real-world datasets, including social networks, citation networks, and sensor network simulations, demonstrate that RL-SGF outperforms existing algorithms in terms of efficiency and solution quality, making it highly applicable to real-world sparse subsystem discovery scenarios in intelligent sensor networks.

This paper studies the temperature effects on device aging, particularly the random telegraph noise (RTN) degradation and the threshold voltage (Vt) shift in a stacked CMOS image sensor (CIS) caused by hot-carrier stress (HCS). Measurements indicate that both are worse when stressed at lower temperatures. Further, the RTN traps generated by HCS can be deactivated effectively by a subsequent high-temperature annealing at 240 °C for up to 360 min. In contrast, the RTN traps in chips not stressed by hot carriers are essentially unaffected by annealing at the same temperature for the same amount of time. This suggests that the physical structure of the RTN traps caused by process-induced damage (PID) without HCS might be different from that generated by HCS. The exact microscopic nature of the differences between these two kinds of RTN traps is not clear and requires further investigation. This work also suggests that RTN degradation could be a useful indicator for device aging for reliability testing and modeling.

CubeSats are a fundamental tool of space exploration, allowing for the testing of novel ideas that can be upscaled to more efficient satellite systems. This work presents the development and characterisation of an additively manufactured aluminium mechanism designed to enable the self-functionalisation of CubeSat structures through material extrusion metal additive manufacturing, as a foundation for sensor integration. A space-grade AlSi7Mg alloy was selected and prepared as a filament to print a fully functional hinge geometry, aiming to evaluate the feasibility of producing movable metallic components using a low-cost and sustainable extrusion-based process. Produced parts were subjected to debinding and vacuum sintering, achieving a densification above 85% and an average hardness of 52.2 HV. Further characterisation, including micro-computed tomography, X-ray diffraction and dynamic mechanical analysis, was used to assess the microstructural integrity, present phase, and mechanical behaviour of the sintered components. The designed shrinkage-compensated hinge mechanism preserved its rotational mobility after sintering, validating the mechanical inter-locking strategy and the design for additive manufacturing methodology used. The results demonstrate that material extrusion enables the fabrication of lightweight, functional, and integrated aluminium mechanisms suitable for sensor incorporation and actuation in small satellite systems. This proof-of-concept highlights material extrusion as a sustainable and economically viable route for developing intelligent aero-space structures, paving the way for future adaptive and sensor-integrated CubeSat subsystems.

Juggling is a complex motor skill that requires multiple sub-skills and cannot be mastered without extensive practice. Although prior studies have quantified performance differences between novice and expert jugglers, none have attempted to quantitatively decompose these components or model their contribution to juggling performance. This longitudinal study presents a multimodal evaluation system that integrates computer vision, motion capture, and biosensing to quantify three key elements of juggling ability: Sequencing, Prediction, and Accuracy. Twenty beginners completed a 10-day, three-ball juggling experiment combining visuo-haptic virtual reality (VR) and real-world practice, with half training in reduced gravity, previously shown to enhance early-stage motor learning. The fitted Gamma-Log generalized linear model (GLM) indicated that Sequencing is the dominant factor of early skill acquisition, followed by Prediction and Accuracy. This study provides the first computational decomposition of juggling, demonstrates how multiple elements jointly contribute to performance, and results in a principled approach to characterizing motor learning in complex real-world tasks.

As power systems expand and grow smarter, the safe and steady operation of substation equipment has become a prerequisite for grid reliability. In cluttered substation scenes, however, existing deep learning detectors still struggle with small targets, multi-scale feature fusion, and precise localization. To overcome these limitations, we introduce PBZGNet, a defect-detection network that couples a gradual parallel-branch backbone, a zoom-fusion neck, and a global channel-recalibration module. First, BiCoreNet is embedded in the feature extractor: dual-core parallel paths, reversible residual links, and channel recalibration cooperate to mine fault-sensitive cues. Second, cross-scale ZFusion and Concat-CBFuse are dynamically merged so that no scale loses information; a hierarchical composite feature pyramid is then formed, strengthening the representation of both complex objects and tiny flaws. Third, an attention-guided decoupled detection head (ADHead) refines responses to obscured and minute defect patterns. Finally, within the Generalized Focal Loss framework, a quality rating scheme suppresses background interference while distribution regression sharpens the localization of small targets. Across all scales, PBZGNet clearly outperforms YOLOv11. Its lightweight variant, PBZGNet-n, attains 83.9% mAP@50 with only 2.91 M parameters and 7.7 GFLOPs-9.3% above YOLOv11-n. The full PBZGNet surpasses the current best substation model, YOLO-SD, by 7.3% mAP@50, setting a new state of the art (SOTA).