IEEE Journal of Selected Areas in Sensors最新文献

Edge-Assisted Federated Learning for Large Language Models in IoT Sensor Systems 物联网传感器系统中大型语言模型的边缘辅助联邦学习

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-02-10 DOI: 10.1109/JSAS.2026.3663337

Hui Chen;Xingyu Yuan;He Li

{"title":"Edge-Assisted Federated Learning for Large Language Models in IoT Sensor Systems","authors":"Hui Chen;Xingyu Yuan;He Li","doi":"10.1109/JSAS.2026.3663337","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3663337","url":null,"abstract":"Federated learning (FL) enables privacy-preserving collaborative training of large-scale AI models, including large language models (LLMs), and is particularly attractive for Internet of Things (IoT) sensor systems with stringent privacy and resource constraints. However, fine-tuning LLMs in sensor-driven edge environments remains challenging due to limited computation and memory resources on edge devices, as well as heterogeneous and time-varying communication and computational conditions that exacerbate the straggler problem. To address these challenges, we propose hierarchical FL framework driven by deep reinforcement learning for large language model (HRL-FLLM), a hierarchical client–edge–cloud FL framework driven by deep reinforcement learning (DRL). HRL-FLLM allows edge servers to adaptively coordinate LLM fine-tuning by jointly controlling efficiency-enhancing techniques, including low-rank adaptation, quantization, and sparse communication, thereby reducing computational overhead and communication costs while preserving model performance. In this article, we further design a convergence-aware reward function and formulate a training time budget optimization problem, enabling edge servers to learn effective local training time allocation policies under heterogeneous conditions. Extensive experiments show that HRL-FLLM achieves robust convergence and significantly reduces training latency. In large-scale heterogeneous FL scenarios with resource-constrained clients, the proposed framework reduces convergence time by up to 47% compared with conventional cloud-centric FL approaches, demonstrating its effectiveness as a scalable and sustainable edge intelligence solution for IoT sensor systems.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"125-138"},"PeriodicalIF":0.0,"publicationDate":"2026-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11387738","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147440606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine Learning-Based Identification and Localization of Closely Spaced Chipless RFID Tags 基于机器学习的近间隔无芯片RFID标签识别与定位

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-30 DOI: 10.1109/JSAS.2026.3659801

F. Villa-Gonzalez;H. Li;R. Bhattacharyya;Sobhi Alfayoumi;S. E. Sarma

{"title":"Machine Learning-Based Identification and Localization of Closely Spaced Chipless RFID Tags","authors":"F. Villa-Gonzalez;H. Li;R. Bhattacharyya;Sobhi Alfayoumi;S. E. Sarma","doi":"10.1109/JSAS.2026.3659801","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3659801","url":null,"abstract":"We present a machine learning-based method for imaging and resolving multiple closely spaced chipless radio frequency identification (RFID) tags within a read zone. Backscattered signal measurements are acquired via a 2-D raster scan using a directive reader antenna and a trained classifier is used to extract the spectral contributions of the tags at each spatial position. This enables the reconstruction of color-coded probability maps that reflect the likely identity and location of each tag. We demonstrate the ability to automatically identify and localize pairs of chipless RFID tags selected from three tag types with distinct resonance frequencies in the 2–6 GHz range. Our method achieves 97.82% accuracy with magnitude measurements and 100% with phase, with average positional errors of less than 6.4 mm when using phase data, even when the tags are positioned at separations below 0.6<inline-formula><tex-math>$lambda$</tex-math></inline-formula>, where mutual coupling is strong. Current limitations and future directions are also discussed.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"113-124"},"PeriodicalIF":0.0,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11368724","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147299689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Robotic Haircutting Systems: A Survey of Methods, Challenges, and Hair Modeling Insights 机器人理发系统：方法，挑战和头发建模见解的调查

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-15 DOI: 10.1109/JSAS.2026.3654480

Ameer Tamoor Khan;Shuai Li

{"title":"Robotic Haircutting Systems: A Survey of Methods, Challenges, and Hair Modeling Insights","authors":"Ameer Tamoor Khan;Shuai Li","doi":"10.1109/JSAS.2026.3654480","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3654480","url":null,"abstract":"The field of robotic haircutting is rapidly evolving, merging advances in service robotics, computer vision, force sensing, and deep learning. This survey article highlights the growing importance of robotic systems in personal grooming, driven by demographic shifts, technological innovation, and increasing demand for hygienic, consistent services. We review the historical evolution of robotic haircutting technologies, including DIY solutions, academic prototypes, and emerging commercial systems. Core technologies, such as robotic arms, hair modeling, motion planning, and tactile sensing are discussed in detail. Key challenges including hair variability, real-time decision-making, user customization, and safety considerations are analyzed. Furthermore, we explore the transformative role of virtual hairstyle modeling and the integration of multimodal perception and AI-driven personalization. Future research directions emphasize the incorporation of large language models, vision-language models, reinforcement learning, and diffusion models to advance human–robot collaboration in salons, while addressing ethical, societal, and inclusivity considerations.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"104-112"},"PeriodicalIF":0.0,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11353904","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Field Programmable Neural Array for Biomedical Signal Processing 生物医学信号处理的现场可编程神经阵列

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-12 DOI: 10.1109/JSAS.2026.3652590

Ingo Hoyer;Raphael Gaede;Alexander Utz;Karsten Seidl

{"title":"Field Programmable Neural Array for Biomedical Signal Processing","authors":"Ingo Hoyer;Raphael Gaede;Alexander Utz;Karsten Seidl","doi":"10.1109/JSAS.2026.3652590","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3652590","url":null,"abstract":"Processing biomedical signals using artificial intelligence (AI) algorithms, particularly convolutional neural networks (CNNs), presents challenges for resource-limited systems, such as smart patches. For instance, screening electrocardiogram (ECG) signals for heart arrhythmias, including atrial fibrillation, is crucial for enhancing early diagnosis, as this condition is closely linked to an increased risk of stroke and overall mortality. Common approaches for System-on-Chip implementation typically involve either a software-based solution utilizing a standard microcontroller or a dedicated hardware accelerator designed to execute the full CNN. This study investigates an approach utilizing reconfigurable fabric, including a theoretical model for efficiency estimation, similar to field programmable gate arrays. These specialized field programmable neural arrays incorporate specialized tiles designed for AI inference. For instance, in the case of a single convolution, the FPNA occupies only half the silicon area while maintaining equivalent timing performance and achieving a 52.9 % reduction in energy consumption per inference. Further results are presented for a test CNN and a CNN for ECG processing. Although the FPNA may require more area and, at times, more energy per inference than a direct hardware implementation, this tradeoff is deemed acceptable due to the advantages of reconfigurability.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"69-77"},"PeriodicalIF":0.0,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11344790","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146082101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Lightweight Algorithm for Detecting Abnormal Interactive Behaviors of Bid Evaluation Personnel Based on Improved YOLOv11n 基于改进YOLOv11n的评标人员异常交互行为轻量检测算法

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-05 DOI: 10.1109/JSAS.2025.3650472

Shanshan Su;Xinyuan Guo;Mingjiang Zhao;Jiaheng Liu;Jiayi Li

{"title":"A Lightweight Algorithm for Detecting Abnormal Interactive Behaviors of Bid Evaluation Personnel Based on Improved YOLOv11n","authors":"Shanshan Su;Xinyuan Guo;Mingjiang Zhao;Jiaheng Liu;Jiayi Li","doi":"10.1109/JSAS.2025.3650472","DOIUrl":"https://doi.org/10.1109/JSAS.2025.3650472","url":null,"abstract":"The intelligent monitoring and management of bid evaluation centers are crucial for enhancing the supervision intensity and level of bidding and bid evaluation work. With the increase in the number of bid evaluation projects and the growing diversification of bid evaluation methods, the introduction of advanced artificial intelligence detection technology has become imperative. A lightweight algorithm is proposed in this study for detecting abnormal interactive behaviors of bid evaluation personnel. The algorithm, derived from an improved version of YOLOv11n, replaces the original model's backbone network with EfficientNet to reduce network scale. In addition, the model incorporates the large kernel separable attention module and bidirectional feature pyramid network (BiFPN) net to enhance detection capacity. Experimental results demonstrate significant improvements in detection accuracy: precision increases by 3.2%, mAP by 1.2%, and F1-score by 1.6%. The improved YOLOv11n model is 3.9 MB in size, 63.9% smaller than the original, with a faster detection time per image by 0.2 ms and reduced FLOPs by 2.4G. These results indicate the model's strong detection capabilities and lightweight advantages, offering potential for efficient bid evaluation supervision support.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"90-103"},"PeriodicalIF":0.0,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11328828","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175756","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A SnO2 Membrane Gas Sensor for Noninvasive Monitoring of Breath Ammonia in Hemodialysis Patients SnO2膜气体传感器无创监测血液透析患者呼吸氨

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-01 Epub Date: 2026-03-02 DOI: 10.1109/JSAS.2026.3668146

Ray-Hua Horng;Hsin-Yu Chou;Dong-Sing Wuu;Chih-Hung Lin;Shang-Feng Tsai;Jia Ning Syu;Cheng-Hsu Chen

{"title":"A SnO2 Membrane Gas Sensor for Noninvasive Monitoring of Breath Ammonia in Hemodialysis Patients","authors":"Ray-Hua Horng;Hsin-Yu Chou;Dong-Sing Wuu;Chih-Hung Lin;Shang-Feng Tsai;Jia Ning Syu;Cheng-Hsu Chen","doi":"10.1109/JSAS.2026.3668146","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3668146","url":null,"abstract":"Breath ammonia is recognized as a potential noninvasive biomarker for renal function assessment. In this study, we applied our previously developed Tin oxide (SnO2) gas sensing technology—to a new clinical application: real-time monitoring of exhaled ammonia in hemodialysis (HD) patients. The SnO2 membrane gas sensor exhibited high sensitivity, low detection limit, and excellent selectivity toward ammonia due to the synergistic effects of the SnO2 membrane. A total of 88 HD patients were recruited, and breath samples were collected immediately before and after dialysis. Results showed a significant decrease in breath ammonia concentration postdialysis (56.02 ± 65.72 ppm to 33.13 ± 39.57 ppm, <italic>p = 0.0021). No significant correlations were found between ammonia changes and conventional dialysis indicators (Kt/V, URR, and nPCR), likely due to sampling time variability and calibration differences. This work demonstrates the potential medical application of a SnO2-based gas sensing platform, providing a rapid and noninvasive approach that may be relevant to bedside dialysis monitoring.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"152-158"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11417826","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147606214","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Edge Imaging With a Low-Area, Low-Power 40 MS/s 0.001 mm$^{2}$ 5-Bit SAR 边缘成像与低面积，低功耗40 MS/s 0.001 mm$^{2}$ 5位SAR

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-01 Epub Date: 2026-03-30 DOI: 10.1109/JSAS.2026.3678515

Sanchari Das;Soumyajit Mandal;Bhasker Choubey;Bibhu Datta Sahoo

{"title":"Edge Imaging With a Low-Area, Low-Power 40 MS/s 0.001 mm$^{2}$ 5-Bit SAR","authors":"Sanchari Das;Soumyajit Mandal;Bhasker Choubey;Bibhu Datta Sahoo","doi":"10.1109/JSAS.2026.3678515","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3678515","url":null,"abstract":"Image sensors designed for edge applications are constrained by power requirements. Furthermore, a number of machine learning tasks can be implemented even with lower bit depth of images, thereby further reducing power requirements in conversion and transmission. Hence, classical 8 or more bit of column parallel analogue-to-digital converters (ADCs) may not be necessarily required in edge applications. In this work, we showcase this potential by designing a 5-bit successive approximation register (SAR) ADC with a sample rate of 40 MS/s while consuming of 380 <inline-formula><tex-math>$mu$</tex-math></inline-formula>W, an active area of 0.001 mm<inline-formula><tex-math>$^{2}$</tex-math></inline-formula> designed in a 65 nm complementary metal–oxide–semiconductor technology. The ADC achieves a signal-to-noise and distortion ratio of 25 dB, a peak signal-to-noise ratio (PSNR) of 32 dB, and a structural similarity index measure of 0.96 when digitizing the canadian institute for advanced research - 10 classes (CIFAR10) dataset. The ADC is also used for image classification tasks on modified national institute of standards and technology (MNIST), Fashion MNIST, extended modified national institute of standards and technology (EMNIST), CIFAR10, ImageNet, DIV2K, and labeled faces in the wild datasets. The accuracy of the digitized images closely matches the accuracy of the original images for all the datsets. The PSNR and the accuracy values show the effectiveness of the proposed ADC architecture for low-power and low-area machine vision applications.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"171-179"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11457606","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147737112","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IP-UNet: Intensity Projection UNet Architecture for IoMT-Enabled 3-D Medical Volume Segmentation IP-UNet：用于支持iomt的三维医疗体分割的强度投影UNet体系结构

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-01 Epub Date: 2026-03-31 DOI: 10.1109/JSAS.2026.3679493

Nyothiri Aung;Mohand Tahar Kechadi;Liming Chen;Sahraoui Dhelim

{"title":"IP-UNet: Intensity Projection UNet Architecture for IoMT-Enabled 3-D Medical Volume Segmentation","authors":"Nyothiri Aung;Mohand Tahar Kechadi;Liming Chen;Sahraoui Dhelim","doi":"10.1109/JSAS.2026.3679493","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3679493","url":null,"abstract":"The proliferation of high-resolution medical imaging data, increasingly generated and transmitted via Internet of Medical Things (IoMT) infrastructure, presents significant computational challenges. Limited memory capacity remains a common drawback when processing these high-resolution 3-D volumetric datasets, as direct cropping or down-sizing often results in a loss of critical diagnostic detail, exacerbates class imbalance, and degrades segmentation performance. 3-D volumes are usually cropped or downsized first before processing, which can result in a loss of resolution, increase class imbalance, and affect the performance of the segmentation algorithms. In this article, we propose an end-to-end deep learning approach called IP-UNet. IP-UNet is a UNet-based model that performs multiclass segmentation on intensity projection (IP) of 3-D volumetric data instead of the memory-consuming 3-D volumes. IP-UNet uses limited memory capability for training without losing the original 3-D image resolution. We compare the performance of three models in terms of segmentation accuracy and computational cost. 1) Slice-by-slice 2-D segmentation of the CT scan images using a conventional 2-D UNet model. 2) IP-UNet that operates on data obtained by merging the extracted maximum intensity projection (MIP), closest vessel projection (CVP), and average intensity projection (AvgIP) representations of the source 3-D volumes, then applying the UNet model on the output IP images. 3) 3-D-UNet model directly reads the 3-D volumes constructed from a series of CT scan images and outputs the 3-D volume of the predicted segmentation. We test the performance of these methods on 3-D volumetric images. Experimental results show that IP-UNet can achieve similar segmentation accuracy with 3-D-UNet but with much better performance. It reduces the training time by 70% and memory consumption by 92%.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"191-198"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11458901","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147796176","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Context-Aware Service Recommendation System for the Social Internet of Things 面向社交物联网的情境感知服务推荐系统

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-01 Epub Date: 2026-02-19 DOI: 10.1109/JSAS.2026.3666177

Amar Khelloufi;Huansheng Ning;Guan Mingxiang;Muhammad Sadiq;Sahraoui Dhelim

{"title":"A Context-Aware Service Recommendation System for the Social Internet of Things","authors":"Amar Khelloufi;Huansheng Ning;Guan Mingxiang;Muhammad Sadiq;Sahraoui Dhelim","doi":"10.1109/JSAS.2026.3666177","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3666177","url":null,"abstract":"The Social Internet of Things (SIoT) refers to the interconnection of Internet of Things (IoT) devices with the addition of a social layer, enabling them to mimic human-like behavior. This technology facet facilitates data and service sharing among devices, creating a scalable network of socially interconnected devices. However, this rapid increase in device numbers and generated data has led to the challenge of data explosion. To address this challenge, service recommendation systems play a crucial role in improving network navigability and service discovery within the SIoT context. However, existing research often overlooks important aspects that could make service selection more pertinent and relevant. Specifically, current service recommendation approaches focus on extracting social relationships between devices while neglecting the contextual presentation of service reviews. In this research, we introduce a service recommendation framework for SIoT context that integrates service-review aggregation and feature learning processes by utilizing the factorization machines to capture complex feature interactions unique to each SIoT device–service pair. Empirical results across three benchmark datasets demonstrate that the proposed Context-Aware Service Recommendation (CASR-SIoT) framework consistently outperforms state-of-the-art baselines, achieving up to 14% improvement in Recall, 8% in Precision, and 15% reduction in RMSE, thereby validating its effectiveness in delivering accurate and context-aware service recommendations in SIoT environments.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"139-151"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11399623","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147557812","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Generative AI-Native Edge Sensors: Collaborative Data Reconstruction via Lightweight Federated Learning 生成式人工智能原生边缘传感器：基于轻量级联邦学习的协同数据重建

IEEE Journal of Selected Areas in Sensors Pub Date : 2026-01-01 Epub Date: 2026-04-02 DOI: 10.1109/JSAS.2026.3680380

Jiaming Pei;Minxi Feng;Pounambal Muthukumar;Parimala Venkata Krishna;Minghui Dai

{"title":"Generative AI-Native Edge Sensors: Collaborative Data Reconstruction via Lightweight Federated Learning","authors":"Jiaming Pei;Minxi Feng;Pounambal Muthukumar;Parimala Venkata Krishna;Minghui Dai","doi":"10.1109/JSAS.2026.3680380","DOIUrl":"https://doi.org/10.1109/JSAS.2026.3680380","url":null,"abstract":"The reliability of time-series data streams in distributed Internet of Things (IoT) sensor networks is frequently compromised by sensor faults and unstable communication, leading to critical data gaps. Traditional centralized approaches for data reconstruction introduce high latency and data privacy risks, failing to meet the requirements of resource-constrained edge environments. To address this, we propose a novel <italic>lightweight federated generative learning framework, termed <italic>FedGen-Sensor, which enables collaborative, privacy-preserving, and resilient data reconstruction directly on heterogeneous edge sensor nodes. Our framework employs a lightweight variational autoencoder optimized for temporal data and integrates a deep <italic>hardware–software co-design approach. Specifically, we utilize an aggressive compression strategy combining <italic>INT8 posttraining quantization and <italic>sparse update transmission to drastically reduce the model’s footprint. Validated on a physical testbed comprising ESP32 microcontrollers and Raspberry Pi servers, FedGen-Sensor demonstrates superior performance: it achieves data reconstruction fidelity comparable to the standard FP32 FedAvg baseline (<inline-formula><tex-math>$text{0.086}$</tex-math></inline-formula> versus 0.082), while simultaneously reducing communication overhead by <inline-formula><tex-math>$text{5.5}times$</tex-math></inline-formula> and cutting the average edge inference latency by <inline-formula><tex-math>$text{2.5}times$</tex-math></inline-formula>. This work validates the feasibility of embedding sophisticated generative intelligence directly onto ultra-low-power edge sensors, setting a new paradigm for autonomous and resilient IoT data services.","PeriodicalId":100622,"journal":{"name":"IEEE Journal of Selected Areas in Sensors","volume":"3 ","pages":"180-190"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11473293","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147796101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0