Pub Date : 2026-10-15Epub Date: 2026-09-15DOI: 10.1109/JSEN.2026.3727638
Chengchi Li;Bao Song;Hu Li;Da Jiang;Yuanlong Xie;Shaowu Lu
Robust 3-D localization and tracking of a noncooperative uncrewed aerial vehicle (UAV) is fundamentally constrained by the tradeoff between wide field-of-view (FoV) coverage and high-resolution, long-range 3-D sensing. Although solid-state light detection and ranging (LiDAR) sensors provide high-resolution and long-range sensing, the narrow FoV often leads to target loss during large-amplitude UAV maneuvers. To mitigate this limitation, this article develops an active sensing framework that integrates a solid-state LiDAR and a monocular camera on a 2-DOF gimbal for continuous single-target UAV localization and tracking, but the dynamic observation introduces two major challenges: rapid gimbal motion causes severe intraframe point-cloud distortion, and long-range sensing yields sparse LiDAR returns together with fluctuating visual detections, making data association unstable. To address these challenges, we propose pointwise LiDAR motion compensation based on tightly coupled inertial measurement unit (IMU) and encoder fusion, 3-D candidate generation with interacting multiple model (IMM)-based motion-frustum dual priors, and a probabilistic data association filter with soft visual confidence and an empty-association hypothesis. Experiments including distortion compensation tests and 16 outdoor UAV flights show effective compensation of gimbal-induced distortion, mean per-flight 3-D RMSE/P95 values of 0.528/0.849 m on confirmed-valid online outputs, and continuous tracking over evaluated ranges of 9.94–51.55 m with the current hardware. The results indicate that active gimbal steering expands the effective observation sector beyond the instantaneous FoV of the LiDAR.
{"title":"Active Localization of Noncooperative UAV Based on Gimbaled LiDAR--Camera Fusion","authors":"Chengchi Li;Bao Song;Hu Li;Da Jiang;Yuanlong Xie;Shaowu Lu","doi":"10.1109/JSEN.2026.3727638","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3727638","url":null,"abstract":"Robust 3-D localization and tracking of a noncooperative uncrewed aerial vehicle (UAV) is fundamentally constrained by the tradeoff between wide field-of-view (FoV) coverage and high-resolution, long-range 3-D sensing. Although solid-state light detection and ranging (LiDAR) sensors provide high-resolution and long-range sensing, the narrow FoV often leads to target loss during large-amplitude UAV maneuvers. To mitigate this limitation, this article develops an active sensing framework that integrates a solid-state LiDAR and a monocular camera on a 2-DOF gimbal for continuous single-target UAV localization and tracking, but the dynamic observation introduces two major challenges: rapid gimbal motion causes severe intraframe point-cloud distortion, and long-range sensing yields sparse LiDAR returns together with fluctuating visual detections, making data association unstable. To address these challenges, we propose pointwise LiDAR motion compensation based on tightly coupled inertial measurement unit (IMU) and encoder fusion, 3-D candidate generation with interacting multiple model (IMM)-based motion-frustum dual priors, and a probabilistic data association filter with soft visual confidence and an empty-association hypothesis. Experiments including distortion compensation tests and 16 outdoor UAV flights show effective compensation of gimbal-induced distortion, mean per-flight 3-D RMSE/P95 values of 0.528/0.849 m on confirmed-valid online outputs, and continuous tracking over evaluated ranges of 9.94–51.55 m with the current hardware. The results indicate that active gimbal steering expands the effective observation sector beyond the instantaneous FoV of the LiDAR.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"31341-31354"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176009","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-08-10DOI: 10.1109/JSEN.2026.3720290
Adong Li;Xiguang Zhang;Junbao Zhang
In the field of ubiquitous acoustic sensing, gesture recognition faces severe performance degradation when generalizing to novel classes under heterogeneous sensor variations and data imbalance. Deep learning models often struggle to distinguish motion-relevant acoustic patterns from device/domain-related nuisance factors. We propose cross-modal physics-aware disentanglement (CPAD), a framework that leverages large language model (LLM)-generated structured semantic descriptions as topology-shaping anchors, rather than explicit physical simulation, to encourage the separation of motion-related features from device/domain-related nuisance factors. To overcome the capacity saturation caused by narrow bottlenecks, CPAD employs an auxiliary projection head paradigm. The dual-branch orthogonality and physics-alignment constraints are utilized strictly as regularizers during pretraining. During inference, these projection heads are discarded, and metric matching is performed directly on the regularized backbone features. Experiments on the ISAC-Gesture dataset show that CPAD achieves a 7.28% absolute improvement over a strictly fair baseline under the dataset-defined unseen-device protocol, reaching 76.25% average accuracy in five-shot scenarios on the available Xiaomi target-device split. For deployment-oriented validation, we report an Android profiling protocol with explicitly stated measurement scopes, covering ONNX export, INT8 quantization, isolated forward-pass latency, few-shot timing, and BatteryManager-based power estimation. INT8 achieves approximately 2.44–$2.46times $ isolated ONNX backbone forward-pass acceleration and reduces the model footprint from 44.7 to 11.2 MB. On vivo X200 Pro mini (V2419A), 25 independent 180-s sustained neural-inference runs show lower mean latency (from 32.78 to 16.89 ms), higher throughput (from 30.33 to 58.82 infer/s), lower average whole-device power estimate (from 6448.74 to 4711.42 mW), and lower normalized per-inference energy (from 212.60 to 80.10 mJ). These results indicate that LLM-structured semantic descriptions can serve as topology-shaping anchors to mitigate sensor-induced degradation under the evaluated acoustic sensing protocol.
在泛在声传感领域，手势识别在异构传感器变化和数据不平衡的情况下，在推广到新类别时面临严重的性能下降。深度学习模型通常难以区分与运动相关的声学模式和与设备/领域相关的干扰因素。我们提出了跨模态物理感知解纠缠（CPAD），这是一个利用大型语言模型（LLM）生成的结构化语义描述作为拓扑塑造锚的框架，而不是明确的物理模拟，以鼓励将运动相关特征与设备/领域相关的干扰因素分离。为了克服窄瓶颈导致的容量饱和，CPAD采用了辅助投影头模式。在预训练过程中严格使用双分支正交性和物理对齐约束作为正则化器。在推理过程中，这些投影头被丢弃，直接对正则化的骨干特征进行度量匹配。在ISAC-Gesture数据集上的实验表明，在数据集定义的隐形设备协议下，CPAD在严格公平的基线上实现了7.28%的绝对改进，在可用的小米目标设备分割的五次射击场景下达到了76.25%的平均准确率。对于面向部署的验证，我们报告了一个带有明确声明测量范围的Android分析协议，涵盖ONNX导出、INT8量化、隔离的前向传递延迟、少射定时和基于batterymanager的功率估计。INT8实现了大约2.44 - $2.46times $隔离ONNX主干前向加速，并将模型占用空间从44.7 MB减少到11.2 MB。在vivo X200 Pro mini （V2419A）上，25个独立的180秒持续神经推理运行显示出更低的平均延迟（从32.78到16.89 ms），更高的吞吐量（从30.33到58.82推断/s），更低的平均整机功率估计（从6448.74到4711.42 mW），更低的标准化推理能量（从212.60到80.10 mJ）。这些结果表明，在评估的声学传感协议下，llm结构的语义描述可以作为拓扑塑造锚点来减轻传感器引起的退化。
{"title":"CPAD: Cross-Modal Physics-Aware Disentanglement for Robust Acoustic Sensing in Gesture Recognition","authors":"Adong Li;Xiguang Zhang;Junbao Zhang","doi":"10.1109/JSEN.2026.3720290","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3720290","url":null,"abstract":"In the field of ubiquitous acoustic sensing, gesture recognition faces severe performance degradation when generalizing to novel classes under heterogeneous sensor variations and data imbalance. Deep learning models often struggle to distinguish motion-relevant acoustic patterns from device/domain-related nuisance factors. We propose cross-modal physics-aware disentanglement (CPAD), a framework that leverages large language model (LLM)-generated structured semantic descriptions as topology-shaping anchors, rather than explicit physical simulation, to encourage the separation of motion-related features from device/domain-related nuisance factors. To overcome the capacity saturation caused by narrow bottlenecks, CPAD employs an auxiliary projection head paradigm. The dual-branch orthogonality and physics-alignment constraints are utilized strictly as regularizers during pretraining. During inference, these projection heads are discarded, and metric matching is performed directly on the regularized backbone features. Experiments on the ISAC-Gesture dataset show that CPAD achieves a 7.28% absolute improvement over a strictly fair baseline under the dataset-defined unseen-device protocol, reaching 76.25% average accuracy in five-shot scenarios on the available Xiaomi target-device split. For deployment-oriented validation, we report an Android profiling protocol with explicitly stated measurement scopes, covering ONNX export, INT8 quantization, isolated forward-pass latency, few-shot timing, and BatteryManager-based power estimation. INT8 achieves approximately 2.44–<inline-formula> <tex-math>$2.46times $ </tex-math></inline-formula> isolated ONNX backbone forward-pass acceleration and reduces the model footprint from 44.7 to 11.2 MB. On vivo X200 Pro mini (V2419A), 25 independent 180-s sustained neural-inference runs show lower mean latency (from 32.78 to 16.89 ms), higher throughput (from 30.33 to 58.82 infer/s), lower average whole-device power estimate (from 6448.74 to 4711.42 mW), and lower normalized per-inference energy (from 212.60 to 80.10 mJ). These results indicate that LLM-structured semantic descriptions can serve as topology-shaping anchors to mitigate sensor-induced degradation under the evaluated acoustic sensing protocol.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30721-30736"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176030","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-08-14DOI: 10.1109/JSEN.2026.3721496
Muhammad Amir Khalil;Shimaa A. Hussien;Mandeep S. Jit Singh;Mohamad A. Alawad;Abdulmajeed M. Alenezi;Mohammad Tariqul Islam
This study presents a compact single-layer circular cross-shaped (CCSS) metamaterial absorber (MTMA) for biomedical refractive-index (RI) sensing applications. The proposed structure, fabricated on an FR-4 substrate with a continuous copper ground plane, achieves near-unity absorption at 4.296 GHz. The absorption response near unity is achieved by simultaneously exciting electric and magnetic resonances, resulting in effective impedance matching at the operating frequency. The design was analyzed using full-wave simulations in Computer Simulation Technology (CST) Studio Suite and further validated through an equivalent circuit model in Advanced Design System, with both approaches producing closely matching results. Due to its symmetric geometry, the absorber is polarization insensitive under normal incidence and preserves absorptivity above 90% for both transverse electric and magnetic polarizations up to an incident angle of 80°. A fabricated prototype was experimentally characterized, and the measured results were in good agreement with the simulated response. For sensing analysis, the resonance frequency shifted by 0.08 GHz per RIU, yielding a sensitivity of 1.12 GHz/RIU and a figure of merit (FoM) of 172. These findings indicate that the proposed absorber is a promising candidate for compact and reliable microwave biomedical sensing applications.
本研究提出了一种紧凑的单层圆形十字形（CCSS）超材料吸收体（MTMA），用于生物医学折射率（RI）传感应用。该结构在具有连续铜接平面的FR-4衬底上制造，在4.296 GHz处实现近均匀吸收。吸收响应接近统一是通过同时激发电共振和磁共振来实现的，从而在工作频率上产生有效的阻抗匹配。采用计算机仿真技术（CST） Studio Suite中的全波仿真分析了该设计，并通过Advanced design System中的等效电路模型进一步验证了该设计，两种方法都产生了紧密匹配的结果。由于其对称的几何结构，吸收体在法向入射下对极化不敏感，并且在入射角度为80°的横向电极化和磁极化下保持90%以上的吸收率。对制作的样机进行了实验表征，测量结果与仿真响应吻合较好。在传感分析中，共振频率每RIU移动0.08 GHz，产生1.12 GHz/RIU的灵敏度和172的品质系数（FoM）。这些发现表明，所提出的吸收剂是一个有希望的紧凑和可靠的微波生物医学传感应用的候选者。
{"title":"Circular Cross-Shaped Metamaterial Absorber With Near-Unity Absorption for Biomedical Microwave Sensing","authors":"Muhammad Amir Khalil;Shimaa A. Hussien;Mandeep S. Jit Singh;Mohamad A. Alawad;Abdulmajeed M. Alenezi;Mohammad Tariqul Islam","doi":"10.1109/JSEN.2026.3721496","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3721496","url":null,"abstract":"This study presents a compact single-layer circular cross-shaped (CCSS) metamaterial absorber (MTMA) for biomedical refractive-index (RI) sensing applications. The proposed structure, fabricated on an FR-4 substrate with a continuous copper ground plane, achieves near-unity absorption at 4.296 GHz. The absorption response near unity is achieved by simultaneously exciting electric and magnetic resonances, resulting in effective impedance matching at the operating frequency. The design was analyzed using full-wave simulations in Computer Simulation Technology (CST) Studio Suite and further validated through an equivalent circuit model in Advanced Design System, with both approaches producing closely matching results. Due to its symmetric geometry, the absorber is polarization insensitive under normal incidence and preserves absorptivity above 90% for both transverse electric and magnetic polarizations up to an incident angle of 80°. A fabricated prototype was experimentally characterized, and the measured results were in good agreement with the simulated response. For sensing analysis, the resonance frequency shifted by 0.08 GHz per RIU, yielding a sensitivity of 1.12 GHz/RIU and a figure of merit (FoM) of 172. These findings indicate that the proposed absorber is a promising candidate for compact and reliable microwave biomedical sensing applications.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30328-30334"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-09-01DOI: 10.1109/JSEN.2026.3726894
Pengxiang Ye;Xiaohan Liu;Hao Wang;Zhaoyuan Tan;Bo Xie;Yulan Lu;Deyong Chen;Junbo Wang
This work presents a high-precision resonant pressure sensor leveraging the volume-compression effect of silicon-glass composite structures for stable ultrahigh-pressure measurement. Different from traditional diaphragm-type sensors, the composite chip undergoes unified integral deformation under hydraulic load to implement pressure detection. We establish a mechanical model for the volume-compression sensing mechanism, laying theoretical support for high-precision measurement under harsh high-temperature high-pressure (HTHP) environments. An Inconel bellows oil-filled packaging structure is adopted to separate the sensing chip from corrosive external fluids and accommodate complex service conditions; a dedicated theoretical model is built to quantify packaging-induced performance variations. The sensing chip is manufactured via standard MEMS micromachining. Packaging comparison tests verify that the bellows packaging reliably isolates media while delivering low-loss pressure transfer. Under $25~^{circ }$ C–$175~^{circ }$ C and 0–172 MPa, experimental calibrations yield a measurement accuracy of 0.025% full scale (FS). High-temperature zero-shift tests confirm outstanding thermal stability, and random vibration & mechanical shock experiments further prove strong environmental robustness. Combining superior pressure resistance, wide thermal tolerance, high precision, and long-term stability, this sensor provides a practical technical solution for ultrahigh-pressure measurement scenarios.
本文提出了一种利用硅-玻璃复合材料结构的体积压缩效应进行稳定超高压测量的高精度谐振压力传感器。与传统的膜片式传感器不同，复合芯片在液压载荷作用下进行统一的积分变形，实现压力检测。建立了体积压缩传感机理的力学模型，为高温高压环境下的高精度测量提供理论支持。采用铬镍铁合金波纹管充油封装结构，使传感芯片与外界腐蚀性流体分离，适应复杂的使用条件；建立了一个专门的理论模型来量化包装引起的性能变化。传感芯片是通过标准的MEMS微加工制造的。封装对比测试验证了波纹管封装在提供低损耗压力传递的同时可靠地隔离介质。在$25~^{circ}$ C - $175~^{circ}$ C和0-172 MPa下，实验校准产生的测量精度为0.025%满量程（FS）。高温零位移试验证实了出色的热稳定性，随机振动和机械冲击实验进一步证明了强大的环境稳健性。该传感器具有优异的耐压性、广泛的热容、高精度和长期稳定性，为超高压测量场景提供了实用的技术解决方案。
{"title":"High-Accuracy MEMS Resonant Sensor Adapted to Extreme High-Temperature and High-Pressure Environments Using Composite Bulk Compression Mechanism","authors":"Pengxiang Ye;Xiaohan Liu;Hao Wang;Zhaoyuan Tan;Bo Xie;Yulan Lu;Deyong Chen;Junbo Wang","doi":"10.1109/JSEN.2026.3726894","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3726894","url":null,"abstract":"This work presents a high-precision resonant pressure sensor leveraging the volume-compression effect of silicon-glass composite structures for stable ultrahigh-pressure measurement. Different from traditional diaphragm-type sensors, the composite chip undergoes unified integral deformation under hydraulic load to implement pressure detection. We establish a mechanical model for the volume-compression sensing mechanism, laying theoretical support for high-precision measurement under harsh high-temperature high-pressure (HTHP) environments. An Inconel bellows oil-filled packaging structure is adopted to separate the sensing chip from corrosive external fluids and accommodate complex service conditions; a dedicated theoretical model is built to quantify packaging-induced performance variations. The sensing chip is manufactured via standard MEMS micromachining. Packaging comparison tests verify that the bellows packaging reliably isolates media while delivering low-loss pressure transfer. Under <inline-formula> <tex-math>$25~^{circ }$ </tex-math></inline-formula>C–<inline-formula> <tex-math>$175~^{circ }$ </tex-math></inline-formula>C and 0–172 MPa, experimental calibrations yield a measurement accuracy of 0.025% full scale (FS). High-temperature zero-shift tests confirm outstanding thermal stability, and random vibration & mechanical shock experiments further prove strong environmental robustness. Combining superior pressure resistance, wide thermal tolerance, high precision, and long-term stability, this sensor provides a practical technical solution for ultrahigh-pressure measurement scenarios.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30409-30417"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176059","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-08-27DOI: 10.1109/JSEN.2026.3726458
Hakeim Talleb
Capacitive and impedimetric lab-on-chip biosensors are designed with field solvers whose discretization error is unknown, so neither the predicted response nor an electrode ranking is guaranteed. We develop a certified-bound framework in which a passive dual transmission-line matrix solver brackets the solution of the governing elliptic problem two-sided through a complementary Prager–Synge pair, a conforming potential field, and an equilibrated flux field. Because the capacitance and the conductance of a cell are the same elliptic problem with a different coefficient, one bracket certifies the geometric factor from which both follow, and this article reports the frequency window in which each is the observable: for phosphate-buffered saline the double layer masks the geometry below 61 kHz and the medium ceases to be a dielectric above 369 MHz, so the certified read-out in a physiological buffer is the conductance. On a dimensioned reference chip the geometric factor is certified to 0.05%, and the framework then bounds what actually limits the design: the fabrication tolerance (1.4% at 100 nm, by conductor monotonicity and with no distributional assumption), the homogenization model error (32%–39% on a recovered concentration, Maxwell–Garnett being shown to coincide with an extremal Hashin–Shtrikman bound rather than a central estimate), the measurement floor and the thermal drift. Ordering that budget shows the numerical certificate to be its smallest term by three orders of magnitude, which is the point: a two-sided numerical bound is what licenses attributing the rest to physics. The same certificate ranks fabricable single-mask electrode patterns in genuine three dimensions, bounds the Hele–Shaw conductance of a meander, and converts Joule heating into a certified read-out drift. Two independent certified arguments, a baseline budget and a thermal budget, converge on the same design rule.
{"title":"Certified-Bound Design of Capacitive and Conductimetric Lab-on-Chip Biosensors: Guaranteed Geometric Factors, Detection Windows, and Electrode Rankings","authors":"Hakeim Talleb","doi":"10.1109/JSEN.2026.3726458","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3726458","url":null,"abstract":"Capacitive and impedimetric lab-on-chip biosensors are designed with field solvers whose discretization error is unknown, so neither the predicted response nor an electrode ranking is guaranteed. We develop a certified-bound framework in which a passive dual transmission-line matrix solver brackets the solution of the governing elliptic problem two-sided through a complementary Prager–Synge pair, a conforming potential field, and an equilibrated flux field. Because the capacitance and the conductance of a cell are the same elliptic problem with a different coefficient, one bracket certifies the geometric factor from which both follow, and this article reports the frequency window in which each is the observable: for phosphate-buffered saline the double layer masks the geometry below 61 kHz and the medium ceases to be a dielectric above 369 MHz, so the certified read-out in a physiological buffer is the conductance. On a dimensioned reference chip the geometric factor is certified to 0.05%, and the framework then bounds what actually limits the design: the fabrication tolerance (1.4% at 100 nm, by conductor monotonicity and with no distributional assumption), the homogenization model error (32%–39% on a recovered concentration, Maxwell–Garnett being shown to coincide with an extremal Hashin–Shtrikman bound rather than a central estimate), the measurement floor and the thermal drift. Ordering that budget shows the numerical certificate to be its smallest term by three orders of magnitude, which is the point: a two-sided numerical bound is what licenses attributing the rest to physics. The same certificate ranks fabricable single-mask electrode patterns in genuine three dimensions, bounds the Hele–Shaw conductance of a meander, and converts Joule heating into a certified read-out drift. Two independent certified arguments, a baseline budget and a thermal budget, converge on the same design rule.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30591-30599"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176065","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study developed a wearable gait analysis system to collect gait data from 71 spinocerebellar ataxia type 3 (SCA3) patients and 38 age-, height-, and weight-matched healthy controls (HCs). We aimed to identify key gait parameters distinguishing SCA3 from HC and examine their correlations with clinical ataxia scales. Gait variability parameters, including stride time standard deviation (SD), single support SD, double support SD, and toe-out angle SD, showed high discriminative power (AUC 0.820–0.840) with large effect sizes. These parameters were significantly correlated with the scale for the assessment and rating of ataxia (SARA) and the International Cooperative Ataxia Rating Scale (ICARS) total scores, as well as disease duration. They not only effectively separated SCA3 patients from HC but also stratified SCA3 severity into mild and moderate subgroups based on SARA scores. The wearable smart insole system provides reliable gait measurements that correlate with clinical severity, supporting its potential clinical utility for the objective assessment of SCA3 progression and severity.
{"title":"A Wearable Sensor-Based Quantitative Gait System for Biomarkers Exploring in Spinocerebellar Ataxia Type 3","authors":"Yue Zhang;Wan-Li Zhang;Xia Liu;Wei Lin;Kai-Liang Luo;Zhao-Di Wang;Xi-Chen Wang;Ying-Juan Zhang;Qi-Kui Sun;Ru-Ying Yuan;Jun Ni;Xin-Yuan Chen;Shu-Wan Wang;Shi-Rui Gan","doi":"10.1109/JSEN.2026.3726075","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3726075","url":null,"abstract":"This study developed a wearable gait analysis system to collect gait data from 71 spinocerebellar ataxia type 3 (SCA3) patients and 38 age-, height-, and weight-matched healthy controls (HCs). We aimed to identify key gait parameters distinguishing SCA3 from HC and examine their correlations with clinical ataxia scales. Gait variability parameters, including stride time standard deviation (SD), single support SD, double support SD, and toe-out angle SD, showed high discriminative power (AUC 0.820–0.840) with large effect sizes. These parameters were significantly correlated with the scale for the assessment and rating of ataxia (SARA) and the International Cooperative Ataxia Rating Scale (ICARS) total scores, as well as disease duration. They not only effectively separated SCA3 patients from HC but also stratified SCA3 severity into mild and moderate subgroups based on SARA scores. The wearable smart insole system provides reliable gait measurements that correlate with clinical severity, supporting its potential clinical utility for the objective assessment of SCA3 progression and severity.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"31299-31308"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-09-14DOI: 10.1109/JSEN.2026.3731419
Lucky Adoh;Daniil Yurchenko;David Thompson;Giacomo Squicciarini;Szu Cheng Lai;Kui Yao
Piezoelectric energy harvesting offers a promising approach for converting ambient vibrations into electrical energy for self-powered sensing systems. Among the most widely used transducers are bending-mode cantilever harvesters operating in the d31 mode and compression-mode multilayer stack harvesters operating in the d33 mode. However, direct experimental comparisons under identical transient railway excitation remain limited. This study presents a comparative investigation of a d31 cantilever and a d33 multilayer stack harvester subjected to representative railway crossing excitation reproduced on an electrodynamic shaker using acceleration signals generated by a validated train–track interaction model. The harvesters were evaluated in terms of voltage response, optimal electrical loading, power output, conversion efficiency, and system-level performance. The d31 harvester achieved a maximum average power of 0.105 mW at an optimal load resistance of 10 k$Omega $ , while the d33 harvester produced 0.085 mW at $100~Omega $ . The d31 harvester achieved a higher conversion efficiency of 10.5%, compared with 8.2% for the d33 harvester. However, the d33 harvester exhibited higher energy density, indicating greater energy generation per unit active piezoelectric volume. The harvested energy also powered a wireless sensing node for 19 sensing and transmission cycles, compared with six using stored energy alone. These results demonstrate the feasibility of self-powered railway monitoring and the influence of transducer configuration under transient excitation.
压电能量收集为自供电传感系统将环境振动转化为电能提供了一种很有前途的方法。其中最广泛使用的传感器是在d31模式下工作的弯曲模式悬臂收割机和在d33模式下工作的压缩模式多层堆叠收割机。然而，在相同暂态铁路励磁条件下的直接实验比较仍然有限。本研究采用经验证的列车-轨道相互作用模型产生的加速度信号，对d31悬臂梁和d33多层堆叠收割机进行了比较研究，并在电动振动筛上再现了代表性的铁路交叉激励。根据电压响应、最佳电负载、功率输出、转换效率和系统级性能对收割机进行了评估。d31收割机在最佳负载电阻为10 k $Omega $时的最大平均功率为0.105 mW，而d33收割机在$100~ $ ω $时的最大平均功率为0.085 mW。d31收割机的转换效率为10.5%，而d33收割机的转换效率为8.2%。然而，d33收集器表现出更高的能量密度，这表明单位有源压电体积产生的能量更大。收集的能量还为无线传感节点提供了19个传感和传输周期，而仅使用存储的能量就需要6个周期。这些结果证明了自供电铁路监测的可行性和瞬态励磁下换能器配置的影响。
{"title":"Comparative Experimental Assessment of d31 and d33 Piezoelectric Harvesters for Railway Energy Harvesting","authors":"Lucky Adoh;Daniil Yurchenko;David Thompson;Giacomo Squicciarini;Szu Cheng Lai;Kui Yao","doi":"10.1109/JSEN.2026.3731419","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3731419","url":null,"abstract":"Piezoelectric energy harvesting offers a promising approach for converting ambient vibrations into electrical energy for self-powered sensing systems. Among the most widely used transducers are bending-mode cantilever harvesters operating in the d<sub>31</sub> mode and compression-mode multilayer stack harvesters operating in the d<sub>33</sub> mode. However, direct experimental comparisons under identical transient railway excitation remain limited. This study presents a comparative investigation of a d<sub>31</sub> cantilever and a d<sub>33</sub> multilayer stack harvester subjected to representative railway crossing excitation reproduced on an electrodynamic shaker using acceleration signals generated by a validated train–track interaction model. The harvesters were evaluated in terms of voltage response, optimal electrical loading, power output, conversion efficiency, and system-level performance. The d<sub>31</sub> harvester achieved a maximum average power of 0.105 mW at an optimal load resistance of 10 k<inline-formula> <tex-math>$Omega $ </tex-math></inline-formula>, while the d<sub>33</sub> harvester produced 0.085 mW at <inline-formula> <tex-math>$100~Omega $ </tex-math></inline-formula>. The d<sub>31</sub> harvester achieved a higher conversion efficiency of 10.5%, compared with 8.2% for the d<sub>33</sub> harvester. However, the d<sub>33</sub> harvester exhibited higher energy density, indicating greater energy generation per unit active piezoelectric volume. The harvested energy also powered a wireless sensing node for 19 sensing and transmission cycles, compared with six using stored energy alone. These results demonstrate the feasibility of self-powered railway monitoring and the influence of transducer configuration under transient excitation.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30548-30555"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176083","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-09-11DOI: 10.1109/JSEN.2026.3730976
Shangjun Lin;Ziheng Zhi;Tiesong Wang;Yufan Zhang;Longhui Zhang;Mingzhu Jiang;Lizhen Zeng;Chunyuan Deng;Fangrong Hu
A flexible polyimide (PI)-supported, single-layer Au interconnected dual-ring metasurface, previously demonstrated with bulk refractive-index sensitivities of 220 GHz/RIU in TM mode and 158 GHz/RIU in TE mode, was developed here into an insulin immunosensor. Direct insulin deposition on the unmodified metasurface generated a concentration-dependent but nonselective dielectric response, motivating construction of a carbon nanotube (CNT)-assisted capture interface and an Ab1–insulin–QD@Ab2 sandwich for selective interfacial loading. QD:Ab2 preparation ratios of 2:1, 5:1, and 10:1 were evaluated by DLS, zeta-potential, and terahertz (THz) measurements; 5:1 produced the most concentrated hydrodynamic-size distribution and the largest dual-mode THz response with modest interdevice dispersion. The complete system was evaluated over an insulin concentration range of 1 fg$cdot $ mL–1–100 ng$cdot $ mL–1, with regression-based estimated LoDs of 9.27 fg$cdot $ mL–1 for TM mode and 7.31 fg$cdot $ mL–1 for TE mode. Specificity controls, bending tests, three regeneration cycles, and storage measurements through day 21 further characterized selectivity and operational robustness. These results show that coordinated metasurface and biointerface design provides a practical basis for sensitive, dual-mode THz immunosensing.
{"title":"Dual-Mode Flexible Terahertz Metasurface Immunosensor for Sensitive Detection of Insulin","authors":"Shangjun Lin;Ziheng Zhi;Tiesong Wang;Yufan Zhang;Longhui Zhang;Mingzhu Jiang;Lizhen Zeng;Chunyuan Deng;Fangrong Hu","doi":"10.1109/JSEN.2026.3730976","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3730976","url":null,"abstract":"A flexible polyimide (PI)-supported, single-layer Au interconnected dual-ring metasurface, previously demonstrated with bulk refractive-index sensitivities of 220 GHz/RIU in TM mode and 158 GHz/RIU in TE mode, was developed here into an insulin immunosensor. Direct insulin deposition on the unmodified metasurface generated a concentration-dependent but nonselective dielectric response, motivating construction of a carbon nanotube (CNT)-assisted capture interface and an Ab<sub>1</sub>–insulin–QD@Ab<sub>2</sub> sandwich for selective interfacial loading. QD:Ab<sub>2</sub> preparation ratios of 2:1, 5:1, and 10:1 were evaluated by DLS, zeta-potential, and terahertz (THz) measurements; 5:1 produced the most concentrated hydrodynamic-size distribution and the largest dual-mode THz response with modest interdevice dispersion. The complete system was evaluated over an insulin concentration range of 1 fg<inline-formula> <tex-math>$cdot $ </tex-math></inline-formula>mL<sup>–1</sup>–100 ng<inline-formula> <tex-math>$cdot $ </tex-math></inline-formula>mL<sup>–1</sup>, with regression-based estimated LoDs of 9.27 fg<inline-formula> <tex-math>$cdot $ </tex-math></inline-formula>mL<sup>–1</sup> for TM mode and 7.31 fg<inline-formula> <tex-math>$cdot $ </tex-math></inline-formula>mL<sup>–1</sup> for TE mode. Specificity controls, bending tests, three regeneration cycles, and storage measurements through day 21 further characterized selectivity and operational robustness. These results show that coordinated metasurface and biointerface design provides a practical basis for sensitive, dual-mode THz immunosensing.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30525-30532"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-10-15Epub Date: 2026-09-16DOI: 10.1109/JSEN.2026.3732643
Zhenyu Dong;Ding Wang;Gang Wang;Yaxing Yue
To address the performance degradation of direct position determination (DPD) algorithms caused by the strong coupling of nonuniform noise and array errors in a moving single-station scenario, this article proposes a two-stage deep learning (DL)-based direct localization method. The joint failure mechanism of these nonideal factors in the DPD method based on the multiple signal classification (MUSIC) algorithm, referred to as DPD-MUSIC, is first theoretically analyzed, demonstrating that the systematic position bias induced by subspace perturbation is difficult to effectively eliminate through analytical approaches. Building upon this analysis, the first stage designs a U-Net denoising network integrated with a convolutional block attention module (CBAM), which employs a real-imaginary decoupling strategy and a triple physics-informed loss function to suppress nonuniform noise and correct array errors. The second stage constructs a multiarchitecture fusion localization network that takes the denoised signal and the spatial position spectrum as dual-path inputs, and achieves subgrid multitarget localization through fractional label encoding and an improved binary cross-entropy (BCE) loss. Simulation results demonstrate that the proposed method achieves significantly higher localization accuracy than multiple representative algorithms under both known and unknown target number scenarios, while exhibiting strong robustness against array errors and nonuniform noise.
{"title":"Direct Position Determination Based on a Two-Stage Deep Learning Model Under Nonuniform Noise and Array Errors","authors":"Zhenyu Dong;Ding Wang;Gang Wang;Yaxing Yue","doi":"10.1109/JSEN.2026.3732643","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3732643","url":null,"abstract":"To address the performance degradation of direct position determination (DPD) algorithms caused by the strong coupling of nonuniform noise and array errors in a moving single-station scenario, this article proposes a two-stage deep learning (DL)-based direct localization method. The joint failure mechanism of these nonideal factors in the DPD method based on the multiple signal classification (MUSIC) algorithm, referred to as DPD-MUSIC, is first theoretically analyzed, demonstrating that the systematic position bias induced by subspace perturbation is difficult to effectively eliminate through analytical approaches. Building upon this analysis, the first stage designs a U-Net denoising network integrated with a convolutional block attention module (CBAM), which employs a real-imaginary decoupling strategy and a triple physics-informed loss function to suppress nonuniform noise and correct array errors. The second stage constructs a multiarchitecture fusion localization network that takes the denoised signal and the spatial position spectrum as dual-path inputs, and achieves subgrid multitarget localization through fractional label encoding and an improved binary cross-entropy (BCE) loss. Simulation results demonstrate that the proposed method achieves significantly higher localization accuracy than multiple representative algorithms under both known and unknown target number scenarios, while exhibiting strong robustness against array errors and nonuniform noise.","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"31503-31520"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rapid development of artificial intelligence and the Internet of Things (IoT) has imposed increasingly stringent signal-to-noise ratio (SNR) requirements on micro-electromechanical system (MEMS) capacitive microphones. However, further SNR improvement in conventional single-ended and single-differential architectures remains limited. To address this issue, this article proposes a multiple-backplate and sealed dual-membrane (MBSDM) MEMS capacitive microphone with four variable capacitors for double-differential readout, thereby enhancing sensitivity and SNR. To predict its performance, three models are developed, including two lumped element models (LEMs) and a full finite element model (FEM). In the two LEMs, the membrane is represented by lumped elements, whose acoustic compliance and acoustic mass are determined by analytical derivation and FEM-based extraction, leading to analytical-LEM and FEM-LEM, respectively. The results show that FEM-LEM agrees closely with the full FEM over the entire frequency range, especially in the voice band; the analytical-LEM maintains good accuracy while offering high computational efficiency. Finally, a noise equivalent circuit is constructed to evaluate the noise performance. At a bias voltage of 5 V, the proposed microphone achieves a predicted sensitivity of −35.7 dBV/Pa and an acoustic SNR of 78.4 dB(A).
{"title":"Design and Modeling of a Novel Multiple-Backplate and Sealed Dual-Membrane MEMS Capacitive Microphone","authors":"Yuhao Miao;Bing Zhang;Genlan Rong;Encheng Zhu;Kai Yan;Wenchao Jin;Shaoping Li;Wei Huang;Songyan Chen","doi":"10.1109/JSEN.2026.3726805","DOIUrl":"https://doi.org/10.1109/JSEN.2026.3726805","url":null,"abstract":"Rapid development of artificial intelligence and the Internet of Things (IoT) has imposed increasingly stringent signal-to-noise ratio (SNR) requirements on micro-electromechanical system (MEMS) capacitive microphones. However, further SNR improvement in conventional single-ended and single-differential architectures remains limited. To address this issue, this article proposes a multiple-backplate and sealed dual-membrane (MBSDM) MEMS capacitive microphone with four variable capacitors for double-differential readout, thereby enhancing sensitivity and SNR. To predict its performance, three models are developed, including two lumped element models (LEMs) and a full finite element model (FEM). In the two LEMs, the membrane is represented by lumped elements, whose acoustic compliance and acoustic mass are determined by analytical derivation and FEM-based extraction, leading to analytical-LEM and FEM-LEM, respectively. The results show that FEM-LEM agrees closely with the full FEM over the entire frequency range, especially in the voice band; the analytical-LEM maintains good accuracy while offering high computational efficiency. Finally, a noise equivalent circuit is constructed to evaluate the noise performance. At a bias voltage of 5 V, the proposed microphone achieves a predicted sensitivity of −35.7 dBV/Pa and an acoustic SNR of 78.4 dB(A).","PeriodicalId":447,"journal":{"name":"IEEE Sensors Journal","volume":"26 20","pages":"30153-30163"},"PeriodicalIF":4.5,"publicationDate":"2026-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：