Search Results (15)

Early-exit deep neural networks (DNNs) can reduce edge-inference latency, but abrupt variations in wireless and computing resources can destabilize split-inference policies. This paper proposes a meta-learning-driven adaptive control framework for joint backbone splitting and early-exit routing in MobileViT. The framework formulates multi-exit splitting as a constrained Markov decision process (CMDP) and introduces splitting-aware multi-dimensional adaptive proximal policy optimization (SMAPPO). SMAPPO combines nonlinear quality-of-service (QoS) penalties with topology-aware action masking, while cross-environment meta-initialization supports edge-local adaptation after resource disturbances. Under the stated simulation assumptions, SMAPPO reached the highest performance-index plateau among six methods in a representative 500-episode stationary trace and achieved the lowest normalized total cost across three latency–energy preference settings. Across ten seeds and nine stationary or disturbed scenarios, online SMAPPO achieved a 77.20% measured accuracy and 22.40 mJ of system energy. With an adaptation horizon of $K=14$, SMAPPO yielded a post-disturbance mean latency of 37.68 ms, a QoS-violation rate of 2.24%, and an on-time completion rate of 98.69%. These results indicate that combining meta-initialization, nonlinear constraint shaping, and topology-aware action masking improves stationary optimization and disturbance recovery within the controlled simulator. Full article

Estimating joint torque in athletes provides important information for understanding neuromuscular performance and optimizing high-intensity training. While surface electromyography (sEMG) offers a non-invasive measure of muscle activation, existing methods often struggle to balance prediction accuracy, computational efficiency, and generalizability across different movement conditions. This study proposes a Hierarchical Three-Stage Feature Selection (H3FS) strategy for efficient and stable sEMG-based joint torque estimation. H3FS integrates domain knowledge, statistical and model-driven screening to generate compact, physiologically meaningful feature sets tailored to various application contexts. Using isokinetic movement data from the shoulder, elbow, and knee joints of 39 young athletes, H3FS identified the optimal configuration, ELATE-3, comprising logarithmic mean absolute value (LMAV), waveform length (WL), and Teager–Kaiser energy operator (TKEO). The study also introduces an Integrated Performance Index (IPI) to evaluate prediction accuracy and computational cost. Results indicate that XGBoost with ELATE-3 achieves the best balance between accuracy and real-time performance (average R² = 0.85, RMSE = 16.2, IPI = 0.013). It outperforms both traditional and data-driven feature sets, demonstrating strong generalization and low latency across joints, making it highly suitable for low-latency estimation on the adopted computational platform. The proposed strategy offers a systematic, scalable solution for joint torque estimation, supporting athletic training optimization and neuromuscular assessment, with potential future applications in rehabilitation monitoring. Full article

Indoor air quality has a significant impact on occupant health, comfort, and productivity in residential and commercial indoor environments. This paper proposes an IoT-edge enabled deep–fuzzy hybrid framework for real-time IAQ prediction and adaptive control. The proposed system integrates IoT-based environmental sensing, Temporal Fusion Transformer-based multivariate forecasting, knowledge distillation, edge-deployed Bi-LSTM inference, and Mamdani fuzzy logic control within a unified IAQ management architecture. A composite Comfort Risk Index is introduced to combine environmental parameters and occupant discomfort feedback into a single adaptive control indicator. Experimental evaluation under varying indoor conditions demonstrated strong forecasting performance, with prediction accuracies reaching 96.3% for CO₂ and 95.7% for PM_2.5 prediction, while reducing inference latency from 575 ms to 295 ms. Comparative analysis against baseline threshold-based control strategies further indicated improved comfort stability, smoother actuator behavior, and reduced estimated actuator operating intensity during deployment. The proposed framework also demonstrated resilient operation under simulated sensor-failure conditions while maintaining low computational overhead suitable for resource-constrained IoT-edge environments. Overall, the results indicate that combining lightweight deep learning models with interpretable fuzzy control can provide an effective, scalable, and energy-aware solution for intelligent real-time IAQ optimization in smart indoor environments. Full article

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

Resolving the kinematic redundancy of 7-DoF humanoid arms to generate natural, human-like motions remains a fundamental challenge in biomimetic robotics. This paper presents a hybrid inverse kinematics (IK) framework that learns a pose-dependent redundancy parameter and integrates it into a differential IK solver. Specifically, we employ the stereographic Shoulder–Elbow–Wrist (SEW) angle as a well-conditioned geometric parameterization. This formulation transforms the algorithmic singularity into a unidirectional half-line, which can be oriented outside the typical reachable workspace. To specify the optimal configuration within the self-motion manifold, a motion dataset was collected by teleoperating a humanoid arm via an anthropomorphic wearable exoskeleton. This approach translates operator-specific postural preferences into the robot’s joint space. A lightweight neural network was then trained to learn the mapping from end-effector poses to these operator-specific SEW angles. By incorporating the predicted SEW angle as a dynamic secondary objective in the null space of the primary tracking task, the proposed framework enables natural redundancy resolution while preserving end-effector tracking accuracy. Both simulations and real-robot experiments were conducted to validate the approach. Results show that, compared to the average performance of static fixed-parameter strategies, the proposed method improves the Joint Configuration Quality Index (CQI) by 22.5% and reduces energy costs by 11.3%. Moreover, the sub-millisecond inference latency (0.44 ms) facilitates seamless integration into real-time control pipelines. Full article

In response to the redundant spatio-temporal modeling and insufficient adaptation to dynamic decision-making in eye-tracking interaction of smart home interfaces, a smart home interface eye-tracking response optimization model based on spatio-temporal Transformer and gate control cross-attention is proposed. It adapts the physiological characteristics of eye-tracking jumps through dynamic sparse attention gating to compress computational redundancy and combines multi-objective reinforcement learning attention modulation to construct a closed-loop decision-making mechanism, optimizing interface parameters in real-time. Experiments showed that the model reduced eye-tracking trajectory prediction error by 23.7% compared to advanced benchmarks, increased the success rate of adapting to dynamic mutation scenarios to 89.2%, and controlled performance fluctuations within 2.3% under noise interference. In high-fidelity user testing, the accuracy of cross-task gaze transfer reached 93.4%, the failure rate of glare interference was optimized to 2.4%, and the user cognitive load index was reduced by 27.9%. Its resource consumption and energy consumption were reduced by 26.7% and 44.9%, respectively, while its posture deviation tolerance remained at 3.5°. The sparse spatio-temporal modeling of the spatio-temporal adaptive Transformer module and the enhanced gating mechanism of the hierarchical gated cross-attention module work together to break through the limitations of traditional methods in computational efficiency and dynamic feedback, providing high-precision and low-latency eye-tracking interaction solutions for smart home interface systems, and promoting the practical evolution of personalized human–machine collaborative control. Full article

In offshore wind farms, modular multilevel converters (MMCs) may operate under a deloading condition to accommodate wind-speed volatility and dispatch constraints. Here, deloading is defined as transmitted power < 0.2 pu (scenario S2, low-power non-reversal). Under this condition, submodule capacitor-voltage fault signatures are weak and exhibit strong operating-point-dependent drift, which degrades conventional threshold-based or offline-trained methods. We propose a lightweight switch-level IGBT open-circuit fault localization framework for deloaded MMCs. Wavelet packet decomposition is used to extract time–frequency energy features, and principal component analysis reduces feature dimensionality for lightweight deployment. An enhanced XGBoost model further integrates severity-index weighting to alleviate class imbalance and incremental learning to adapt to condition drift induced by wind-power fluctuations. MATLAB2024b/Simulink results show 99.6% accuracy in S2 with less than 2 ms inference latency, and robust performance in extended scenarios including partial-power operation and power reversal. Full article

Unmanned Aerial Systems (UASs) are playing an increasingly critical role in both civilian and defense applications. However, their heavy reliance on unencrypted Global Navigation Satellite System (GNSS) signals, particularly GPS, makes them highly susceptible to signal spoofing attacks, posing severe operational and safety threats. This paper introduces a comprehensive, AI-driven multi-layer sensor framework that simultaneously enables real-time spoofing detection and secure command-and-control (C2) communication in lightweight UAS platforms. The proposed system enhances telemetry reliability through a refined preprocessing pipeline that includes a novel GPS Drift Index (GDI), robust statistical normalization, cluster-constrained oversampling, Kalman-based noise reduction, and quaternion filtering. These sensing layers improve anomaly separability under adversarial signal manipulation. On this enhanced feature space, a differentiable architecture search (DARTS) approach dynamically generates lightweight neural network architectures optimized for fast, onboard spoofing detection. For secure command and control, the framework integrates a low-latency cryptographic layer utilizing PRESENT-128 encryption and CMAC authentication, achieving confidentiality and integrity with only 1.79 ms latency and a 0.51 mJ energy cost. Extensive experimental evaluations demonstrate the framework’s outstanding detection accuracy (99.99%), near-perfect F1-score (0.999), and AUC (0.9999), validating its suitability for deployment in real-world, resource-constrained UAS environments. This research advances the field of AI-enabled sensor systems by offering a robust, scalable, and secure navigation framework for countering GPS spoofing in autonomous aerial vehicles. Full article

Liquid crystals (LCs) with fluid-like flow and solid-like molecular orientation find important applications in optical display and sensor technologies. Predicting the mean steady-state polar angle and refractive index is crucial for optimizing LC performance. While conventional predictive models such as those based on continuum theories require complex and computationally intensive numerical simulations, this study employs artificial neural networks (ANNs). In particular, they are developed to predict the mean steady state polar angle and refractive index from surface viscosity and anchoring energy. Using the train, validation, test method, ANN_A4 (R² = 0.9995) and ANN_B2 (R² = 0.9969) are found to have the highest predictive accuracy. On the other hand, using the K-Fold cross-validation, the results significantly differ, with the best performance shown in ANN_A5* (R² = 0.40767) and ANN_B4* (R² = 0.93799). Coupled with the low latency of ANNs, these results indicate that ANNs have significant potential in LC modeling, especially for use in the computationally intensive optimization of LC-based technologies. Full article

Sleep is increasingly acknowledged as a cornerstone of public health, with chronic sleep loss implicated in preventable injury and deaths. Obstructive sleep apnea (OSA) affects over one billion people worldwide but remains widely under-diagnosed due to dependence on polysomnography (PSG), an overnight, hospital-based intrusive procedure. As an adjunct to the clinical diagnosis of OSA, this paper presents a low-cost, smartphone-based Generative AI agent framework for sleep apnea detection and sleep coaching at the bedside. Powered by an on=device Generative AI model, the four agents of this framework include a classifier, an analyser, a visualiser, and a sleep coach. The key agent activities performed are sleep apnea detection, sleep data management, data analysis, and natural language sleep coaching. The framework was empirically evaluated on a subject-independent hold-out set drawn from a dataset of 500 clinician annotated clips collected from 10 clinically diagnosed OSA patients. Sleep apnea detection achieved an accuracy of 0.89, precision of 0.91, and recall of 0.88, with nightly Apnea–Hypopnea Index (AHI) estimates strongly correlated with PSG-based clinical scores. The framework was further assessed on the performance metrics of computation, latency, memory, and energy usage. The results of these experiments confirm the feasibility of the proposed framework for large-scale, low-cost OSA screening, with pathways for future work in federated learning, noise robustness, and broad clinical validation. Full article

Underwater object detection is critical for marine resource utilization, ecological monitoring, and maritime security, yet it remains constrained by optical degradation, high energy consumption, and vulnerability to adversarial perturbations. To address these challenges, this study proposes an Adaptive Vision Transformer (A-ViT)-based detection framework. At the hardware level, a systematic power-modeling and endurance-estimation scheme ensures feasibility across shallow- and deep-water missions. Through the super-resolution reconstruction based on the Hybrid Attention Transformer (HAT) and the staged enhancement with the Deep Initialization and Deep Inception and Channel-wise Attention Module (DICAM), the image quality was significantly improved. Specifically, the Peak Signal-to-Noise Ratio (PSNR) increased by 74.8%, and the Structural Similarity Index (SSIM) improved by 375.8%. Furthermore, the Underwater Image Quality Measure (UIQM) rose from 3.00 to 3.85, while the Underwater Color Image Quality Evaluation (UCIQE) increased from 0.550 to 0.673, demonstrating substantial enhancement in both visual fidelity and color consistency. Detection accuracy is further enhanced by an improved YOLOv11-Coordinate Attention–High-order Spatial Feature Pyramid Network (YOLOv11-CA_HSFPN), which attains a mean Average Precision at Intersection over Union 0.5 (mAP@0.5) of 56.2%, exceeding the baseline YOLOv11 by 1.5 percentage points while maintaining 10.5 ms latency. The proposed A-ViT + ROI reduces inference latency by 27.3% and memory usage by 74.6% when integrated with YOLOv11-CA_HSFPN and achieves up to 48.9% latency reduction and 80.0% VRAM savings in other detectors. An additional Image-stage Attack QuickCheck (IAQ) defense module reduces adversarial-attack-induced latency growth by 33–40%, effectively preventing computational overload. Full article

This study presents a novel federated learning (FL) methodology implemented directly on STM32-based microcontrollers (MCUs) for energy-efficient smart irrigation. To the best of our knowledge, this is the first work to demonstrate end-to-end FL training and aggregation on real STM32 MCU clients (STM32F722ZE), under realistic energy and memory constraints. Unlike most prior studies that rely on simulated clients or high-power edge devices, our framework deploys lightweight neural networks trained locally on MCUs and synchronized via message queuing telemetry transport (MQTT) communication. Using a smart agriculture (SA) dataset partitioned by soil type, 7 clients collaboratively trained a model over 3 federated rounds. Experimental results show that MCU clients achieved competitive accuracy (70–82%) compared to PC clients (80–85%) while consuming orders of magnitude less energy. Specifically, MCU inference required only 0.95 mJ per sample versus 60–70 mJ on PCs, and training consumed ∼70 mJ per epoch versus nearly 20 J. Latency remained modest, with MCU inference averaging 3.2 ms per sample compared to sub-millisecond execution on PCs, a negligible overhead in irrigation scenarios. The evaluation also considers the payoff between accuracy, energy consumption, and latency through the Energy Latency Accuracy Index (ELAI). This integrated perspective highlights the trade-offs inherent in deploying FL on heterogeneous devices and demonstrates the efficiency advantages of MCU-based training in energy-constrained smart irrigation settings. Full article

Accurate and timely detection of islanding events is essential for ensuring the stability and safety of hybrid power systems with high penetration of distributed energy resources. Traditional islanding detection methods often face challenges related to detection speed, false alarms, and robustness under dynamic operating conditions. This paper proposes a Robust Random Vector Functional Link Network (RVFLN)-based detection framework that leverages engineered features extracted from voltage, current, and power signals in a hybrid microgrid. The proposed method integrates statistical, spectral, and spatiotemporal features—including the Dynamic Harmonic Profile (DHP), which tracks rapid harmonic distortions during disconnection, the Sub-band Energy Ratio (SBER), which quantifies the redistribution of signal energy across frequency bands, and the Islanding Anomaly Index (IAI), which measures multivariate deviations in system behavior—capturing both transient and steady-state characteristics. A real-time digital simulator (RTDS) is used to model diverse scenarios including grid-connected operation, islanding at the Point of Common Coupling (PCC), synchronous converter islanding, and fault events. The RVFLN is trained and validated using this high-fidelity data, enabling robust classification of operational states. Results demonstrate that the RVFLN achieves high accuracy (up to 98.5%), low detection latency (average 0.05 s), and superior performance across precision, recall, and F1 score compared to conventional classifiers such as Random Forest, SVM, and k-NN. The proposed approach ensures reliable real-time islanding detection, making it a strong candidate for deployment in intelligent protection and monitoring systems in modern power networks. Full article

Researchers have studied instances of power line technical failures, the significant rise in the energy loss index in the line connecting the distribution transformer and consumer meters, and the inability to control unauthorized line connections. New, innovative, and scientific approaches are required to address these issues while enhancing the reliability and efficiency of electricity supply. This study evaluates the reliability of Internet of Things (IoT)-aided remote monitoring systems specifically designed for a low-voltage overhead transmission line. Many methods of analysis and comparison have been employed to examine the reliability of wireless sensor devices used in real-time remote monitoring. A reliability model was developed to evaluate the reliability of the monitoring system in various situations. Based on the developed models, it was found that the reliability indicators of the proposed monitoring system were 98% in 1 month. In addition, it has been proven that the reliability of the system remains high even when an optional sensor in the network fails. This study investigates various IoT technologies, their integration into monitoring systems, and their effectiveness in enhancing the reliability and efficiency of electrical transmission infrastructure. The analysis includes data from field deployments, case studies, and simulations to assess performance metrics, such as accuracy, latency, and fault detection capabilities. Full article

Dynamic skyline query is one of the most popular and significant variants of skyline query in the field of multi-criteria decision-making. However, designing a distributed dynamic skyline query possesses greater challenge, especially for the distributed data centric storage within wireless sensor networks (WSNs). In this paper, a novel Enhanced Distributed Dynamic Skyline (EDDS) approach is proposed and implemented in Disk Based Data Centric Storage (DBDCS) architecture. DBDCS is an adaptation of magnetic disk storage platter consisting tracks and sectors. In DBDCS, the disc track and sector analogy is used to map data locations. A distance based indexing method is used for storing and querying multi-dimensional similar data. EDDS applies a threshold based hierarchical approach, which uses temporal correlation among sectors and sector segments to calculate a dynamic skyline. The efficiency and effectiveness of EDDS has been evaluated in terms of latency, energy consumption and accuracy through a simulation model developed in Castalia. Full article