Search Results (1,527)

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

Fe³⁺-incorporated hydrogels are particularly valuable for wearable devices due to their tunable mechanical properties and ionic conductivity. However, conventional immersion-based fabrication fundamentally limits hydrogel performance because of heterogeneous ion distribution, ionic leaching, and scalability limitations. To overcome these challenges, we report a novel one-pot strategy where controlled amounts of Fe³⁺ are directly added to polyacrylamide-sodium acrylate (PAM-SA) precursor solutions, ensuring homogeneous ion distribution. Combining this with Photoinduced Electron/Energy Transfer Reversible Addition–Fragmentation Chain Transfer (PET-RAFT) polymerization enables efficient hydrogel fabrication under open-vessel conditions, improving its scalability. Fe³⁺ concentration achieves unprecedented modulation of mechanical properties: Young’s modulus (10 to 150 kPa), toughness (0.26 to 2.3 MJ/m³), and strain at break (800% to 2500%). The hydrogels also exhibit excellent compressibility (90% strain recovery), energy dissipation (>90% dissipation efficiency at optimal Fe³⁺ levels), and universal adhesion to diverse surfaces (plastic, metal, PTFE, and cardboard). Finally, these Fe³⁺-incorporated hydrogels demonstrated high effectiveness as strain sensors for monitoring finger/elbow movements, with gauge factors dependent on composition. This work provides a scalable, oxygen-tolerant route to tunable hydrogels for advanced wearable devices. Full article

The Internet of Things (IoT) is transforming how devices communicate, with more than 30 billion connected units today and projections exceeding 40 billion by 2025. Despite this growth, the integration of heterogeneous systems remains a significant challenge, particularly in sensitive domains like healthcare, where proprietary standards and isolated ecosystems hinder interoperability. This paper presents an extended version of the Share design pattern, a lightweight and contract-based mechanism for dynamic service composition, tailored for resource-constrained IoT devices. Share enables decentralized, peer-to-peer integration by exchanging executable code in our examples written in the LUA programming language. This approach avoids reliance on centralized infrastructures and allows services to discover and interact with each other dynamically through pattern-matching and contract validation. To assess its suitability, we developed an emulator that directly implements the system under test in LUA, allowing us to verify both the structural and behavioral constraints of service interactions. Our results demonstrate that Share is scalable and effective, even in constrained environments, and supports formal correctness via design-by-contract principles. This makes it a promising solution for lightweight, interoperable IoT systems that require flexibility, dynamic configuration, and resilience without centralized control. Full article

With the increasing scale of industrial devices and the growing complexity of multi-source heterogeneous sensor data, traditional methods struggle to address challenges in fault detection, data security, and trustworthiness. Ensuring tamper-proof data storage and improving prediction accuracy for imbalanced anomaly detection for potential deployment in the Industrial Internet of Things (IIoT) remain critical issues. This study proposes a blockchain-powered Long Short-Term Memory Network (LSTM)–Attention hybrid model: an LSTM-based Encoder–Attention–Decoder (LEAD) for industrial device anomaly detection. The model utilizes an encoder–attention–decoder architecture for processing multivariate time series data generated by industrial sensors and smart contracts for automated on-chain data verification and tampering alerts. Experiments on real-world datasets demonstrate that the LEAD achieves an $F_{0.1}$ score of 0.96, outperforming baseline models (Recurrent Neural Network (RNN): 0.90; LSTM: 0.94; and Bi-directional LSTM (Bi-LSTM, 0.94)). We simulate the system using a private FISCO-BCOS network with a multi-node setup to demonstrate contract execution, anomaly data upload, and tamper alert triggering. The blockchain system successfully detects unauthorized access and data tampering, offering a scalable solution for device monitoring. Full article

The integration of unmanned aerial vehicles (UAVs) into next-generation wireless networks greatly enhances the flexibility and efficiency of communication and distributed computation for ground mobile devices. Federated learning (FL) provides a privacy-preserving paradigm for device collaboration but remains highly vulnerable to poisoning attacks and is further challenged by the resource constraints and heterogeneous data common to UAV-assisted systems. Existing robust aggregation and anomaly detection methods often degrade in efficiency and reliability under these realistic adversarial and non-IID settings. To bridge these gaps, we propose FedULite, a lightweight and robust federated learning framework specifically designed for UAV-assisted environments. FedULite features unsupervised local representation learning optimized for unlabeled, non-IID data. Moreover, FedULite leverages a robust, adaptive server-side aggregation strategy that uses cosine similarity-based update filtering and dimension-wise adaptive learning rates to neutralize sophisticated data and model poisoning attacks. Extensive experiments across diverse datasets and adversarial scenarios demonstrate that FedULite reduces the attack success rate (ASR) from over 90% in undefended scenarios to below 5%, while maintaining the main task accuracy loss within 2%. Moreover, it introduces negligible computational overhead compared to standard FedAvg, with approximately 7% additional training time. Full article

Simulation is an important technical tool in corn threshing operations, and the establishment of the corn kernel model is the core part of the simulation process. The existing modeling method is to treat the whole kernel as a rigid body, which cannot be crushed during the simulation process, and the calculation of the crushing rate needs to be considered through multiple criteria such as the contact force, the number of collisions, and so on. Aiming at the issue that kernel crushing during maize threshing cannot be accurately modeled in discrete element simulations, in this study, a sub-area crushing model was constructed; representative samples with 26%, 30% and 34% moisture content were selected from a double-season maturing region in China; based on the physical dimensions and biological structure of the maize kernel, three stress regions were defined; and mechanical property tests were conducted on each of the three stress regions using a texturometer as a way to determine the different crushing forces due to the heterogeneity of the maize structure. The correctness of the model was verified by stacking angle and mechanical property experiments. A discrete element model of corn kernels was established using the Bonding V2 method and sub-area modeling. Bonding parameters were calculated by combining stacking angle tests and mechanical property tests. The flattened corn kernel was used as a prototype, and the bonding parameters were determined through size and mechanical property tests. A 22-ball bonding model was developed using dimensional parameters, and the kernel density was recalculated. Results showed that the relative error between the stacking angle test and the measured mean value was 0.31%. The maximum deviation of axial compression simulation results from the measured mean value was 22.8 N, and the minimum deviation was 3.67 N. The errors between simulated and actual rupture forces at the three force areas were 5%, 10%, and 0.6%, respectively. The decreasing trend of the maximum rupture force for the three moisture levels in the simulation matched that of the actual rupture force. The discrete element model can accurately reflect the rupture force, energy relationship, and rupture process on both sides, top, and bottom of the grain, and it can solve the error problem caused by the contact between the threshing element and the grain line in the actual threshing process to achieve the design optimization of the threshing drum. The modeling method provided in this study can also be applied to breakable discrete element models for wheat and soybean, and it provides a reference for optimizing the design of subsequent threshing devices. Full article

The rapid integration of complex sensors and electronic control units (ECUs) in autonomous vehicles significantly increases cybersecurity risks in vehicular networks. Although the Controller Area Network (CAN) is efficient, it lacks inherent security mechanisms and is vulnerable to various network attacks. The traditional Intrusion Detection System (IDS) makes it difficult to effectively deal with the dynamics and complexity of emerging threats. To solve these problems, a lightweight vehicular network intrusion detection framework based on Dynamic Feature Fusion Federated Learning (DFF-FL) is proposed. The proposed framework employs a two-stream architecture, including a transformer-augmented autoencoder for abstract feature extraction and a lightweight CNN-LSTM–Attention model for preserving temporal and local patterns. Compared with the traditional theoretical framework of the federated learning, DFF-FL first dynamically fuses the deep feature representation of each node through the transformer attention module to realize the fine-grained cross-node feature interaction in a heterogeneous data environment, thereby eliminating the performance degradation caused by the difference in feature distribution. Secondly, based on the final loss $L_{AE}\left(X,\widehat{X}\right)$ index of each node, an adaptive weight adjustment mechanism is used to make the nodes with excellent performance dominate the global model update, which significantly improves robustness against complex attacks. Experimental evaluation on the CAN-Hacking dataset shows that the proposed intrusion detection system achieves more than 99% F1 score with only 1.11 MB of memory and 81,863 trainable parameters, while maintaining low computational overheads and ensuring data privacy, which is very suitable for edge device deployment. Full article

This comprehensive review maps the fast-evolving landscape in which artificial intelligence (AI) and deep-learning (DL) techniques converge with the Internet of Things (IoT) to manage energy, comfort, and sustainability across smart environments. A PRISMA-guided search of four databases retrieved 1358 records; after applying inclusion criteria, 143 peer-reviewed studies published between January 2019 and April 2025 were analyzed. This review shows that AI-driven controllers—especially deep-reinforcement-learning agents—deliver median energy savings of 18–35% for HVAC and other major loads, consistently outperforming rule-based and model-predictive baselines. The evidence further reveals a rapid diversification of methods: graph-neural-network models now capture spatial interdependencies in dense sensor grids, federated-learning pilots address data-privacy constraints, and early integrations of large language models hint at natural-language analytics and control interfaces for heterogeneous IoT devices. Yet large-scale deployment remains hindered by fragmented and proprietary datasets, unresolved privacy and cybersecurity risks associated with continuous IoT telemetry, the growing carbon and compute footprints of ever-larger models, and poor interoperability among legacy equipment and modern edge nodes. The authors of researches therefore converges on several priorities: open, high-fidelity benchmarks that marry multivariate IoT sensor data with standardized metadata and occupant feedback; energy-aware, edge-optimized architectures that lower latency and power draw; privacy-centric learning frameworks that satisfy tightening regulations; hybrid physics-informed and explainable models that shorten commissioning time; and digital-twin platforms enriched by language-model reasoning to translate raw telemetry into actionable insights for facility managers and end users. Addressing these gaps will be pivotal to transforming isolated pilots into ubiquitous, trustworthy, and human-centered IoT ecosystems capable of delivering measurable gains in efficiency, resilience, and occupant wellbeing at scale. Full article

Obstructive sleep apnea (OSA) is a prevalent and heterogeneous sleep disorder associated with significant health and societal burdens. While continuous positive airway pressure (CPAP) remains the gold standard treatment, its limitations in adherence and patient tolerance have highlighted the need for alternative therapies. Positional therapy (PT), which targets apneas that occur predominantly in the supine position, has emerged as a promising option for individuals with positional OSA (POSA). This narrative review synthesizes the current literature on PT, examining its clinical indications, typologies, comparative efficacy with CPAP, oral appliances, and hypoglossal nerve stimulation, as well as data on adherence and barriers to long-term use. Traditional methods such as the tennis ball technique have largely been replaced by modern vibrotactile devices, which demonstrate improved comfort, adherence, and comparable short-term outcomes in selected POSA subjects. While PT remains inferior to CPAP in reducing overall AHI and oxygen desaturation, it performs favorably in terms of mean disease alleviation (MDA) and sleep continuity. Importantly, treatment effectiveness is influenced by both anatomical and non-anatomical traits, underscoring the need for accurate phenotyping and individualized care. PT should be considered within a broader patient-centered model that incorporates preferences, lifestyle, and motivational factors. Further research is needed to validate long-term efficacy, optimize selection criteria, and integrate PT into personalized OSA management strategies. Full article

A new, simple machine was developed to address a long-standing challenge in biomedical and mechanical engineering: how to enhance the primary stability and long-term integration of screws and implants in low-density or heterogeneous materials, such as bone or composite substrates. Traditional screws often rely solely on external threading for fixation, leading to limited cohesion, poor integration, or early loosening under cyclic loading. In response to this problem, we designed and built a novel device that leverages a unique mechanical principle to simultaneously perforate, collect, and compact the substrate material during insertion. This mechanism results in an internal material interlock, enhancing cohesion and stability. Drawing upon principles from physics, chemistry, engineering, and biology, we evaluated its biomechanical behavior in synthetic bone analogs. The maximum insertion (MIT) and removal torques (MRT) were measured on synthetic osteoporotic bones using a digital torquemeter, and the values were compared directly. Experimental results demonstrated that removal torque (mean of 21.2 Ncm) consistently exceeded insertion torque (mean of 20.2 Ncm), indicating effective material interlocking and cohesive stabilization. This paper reviews the relevant literature, presents new data, and discusses potential applications in civil infrastructure, aerospace, and energy systems where substrate cohesion is critical. The findings suggest that this new simple machine offers a transformative approach to improving fixation and integration across multiple domains. Full article

Background/Objectives: To review the dimensions of the round window region (round window niche, bony structures surrounding the niche, and the membrane itself). Methods: Medline, EMBASE, Cochrane Library, and Google Scholar databases were searched by two independent reviewers. Anatomical and radiological studies on the round window region were screened. Studies reporting at least one dimension for the round window (RW) niche and/or the RW membrane were included. Results: Sixteen studies met the inclusion criteria (13 anatomical and 3 radiological studies) for a total number of 808 temporal bones with at least one dimension reported. The structures measured varied across the different studies with 12 reporting RW membrane dimensions (area and/or at least one distance), 8 detailing RW niche dimensions (height, width or depth) and 6 which measured at least one element of the RW bony overhangs (posterior or anterior pillar, RW tegmen). Surface area of the RW membrane varied between 0.32 mm² and 2.89 mm², with a minimum dimension (minimum diameter or height or width) comprising between 0.51 mm and 2.1 mm. When the bony overhangs surrounding the membrane were not considered, the minimum diameter was between 1.65 mm and 1.97 mm. Conclusions: The dimensions of the RW region are intrinsically variable, but the heterogeneity of the measurements reported also contributes to these variations. Posterior pillar, RW tegmen, anterior pillar, and their relative development probably account for a large part of this variability. The future RW membrane devices should be ≤1 mm in their maximum dimension, whether or not individually tailored, to fit most of the RW membranes. Full article

Background and Objectives: Postoperative pancreatic fistula and post-hepatectomy liver failure remain significant complications after HPB surgery; however, superficial surgical site infection (SSI) is the most frequent wound-related complication. Closed-incision negative-pressure wound therapy (ciNPWT) has been proposed to reduce superficial contamination, yet no liver-focused quantitative synthesis exists. We aimed to evaluate the effectiveness and safety of prophylactic ciNPWT after hepatopancreatobiliary (HPB) surgery. Methods: MEDLINE, Embase, and PubMed were searched from inception to 30 April 2025. Randomized and comparative observational studies that compared ciNPWT with conventional dressings after elective liver transplantation, hepatectomy, pancreatoduodenectomy, and liver resections were eligible. Two reviewers independently screened, extracted data, and assessed risk of bias (RoB-2/ROBINS-I). A random-effects Mantel–Haenszel model generated pooled risk ratios (RRs) for superficial SSI; secondary outcomes were reported descriptively. Results: Twelve studies (seven RCTs, five cohorts) encompassing 15,212 patients (3561 ciNPWT; 11,651 control) met the inclusion criteria. Device application lasted three to seven days in all trials. The pooled analysis demonstrated a 29% relative reduction in superficial SSI with ciNPWT (RR 0.71, 95% CI 0.63–0.79; p < 0.001) with negligible heterogeneity (I² 0%). Absolute risk reduction ranged from 0% to 13%, correlating positively with the baseline control-group SSI rate. Deep/organ-space SSI (RR 0.93, 95% CI 0.79–1.09) and 90-day mortality (RR 0.94, 95% CI 0.69–1.28) were unaffected. Seven studies documented a 1- to 3-day shorter median length of stay; only two reached statistical significance. Device-related adverse events were rare (one seroma, no skin necrosis). Conclusions: Prophylactic ciNPWT safely reduces superficial SSI after high-risk HPB surgery, with the greatest absolute benefit when baseline SSI risk exceeds ≈10%. Its influence on deep infection and mortality is negligible. Full article

This paper examines the operational consequences and energy demands associated with the growing cultural diversity of air travellers at airport security checkpoints. The analysis focuses on how an increasing proportion of passengers requiring enhanced security screening, due to cultural, religious, or linguistic factors, affects both system throughput and energy consumption. The methodology integrates synchronised measurement of passenger flow with real-time monitoring of electricity usage. Four operational scenarios, representing incremental shares (0–15%) of passengers subject to extended screening, were modelled. The findings indicate that a 15% increase in this passenger group leads to a statistically significant rise in average power consumption per device (3.5%), a total energy usage increase exceeding 4%, and an extension of average service time by 0.6%—the cumulative effect results in a substantial annual contribution to the airport’s carbon footprint. The results also reveal a higher frequency and intensity of power consumption peaks, emphasising the need for advanced infrastructure management. The study emphasises the significance of predictive analytics, dynamic resource allocation, and the implementation of energy-efficient technologies. Furthermore, systematic intercultural competency training is recommended for security staff. These insights provide a scientific basis for optimising airport security operations amid increasing passenger heterogeneity. Full article

The evolution from 4G to 5G—and eventually to the forthcoming 6G networks—has revolutionized wireless communications by enabling high-speed, low-latency services that support a wide range of applications, including the Internet of Things (IoT), smart cities, and critical infrastructures. However, the unique characteristics of these networks—extensive connectivity, device heterogeneity, and architectural flexibility—impose significant security challenges. This paper introduces a comprehensive framework for enhancing the security of current and emerging wireless networks by integrating state-of-the-art machine learning (ML) techniques into intrusion detection and prevention systems. It also thoroughly explores the key aspects of wireless network security, including architectural vulnerabilities in both 5G and future 6G networks, novel ML algorithms tailored to address evolving threats, privacy-preserving mechanisms, and regulatory compliance with the EU AI Act. Finally, a Wireless Intrusion Detection Algorithm (WIDA) is proposed, demonstrating promising results in improving wireless network security. Full article

Transparency is a key requirement for some classes of robots, exoskeletons, and assistive devices (READs), where safe and efficient human–robot interaction is crucial. Typical fields that require transparency are rehabilitation and industrial contexts. However, the definitions of transparency adopted in the literature are heterogeneous. It follows that there is a need to clarify, summarize, and assess how transparency is commonly defined and measured. Thus, the goal of this review is to systematically examine how transparency is conceptualized and evaluated across studies. To this end, we performed a structured search across three major scientific databases. After a thorough screening process, 20 out of 400 identified articles were further examined and included in this review. Despite being recognized as a desirable and essential characteristic of READs in many domains of application, our findings reveal that transparency is still inconsistently defined and evaluated, which limits comparability across studies and hinders the development of standardized evaluation frameworks. Indeed, our screening found significant heterogeneity in both terminology and evaluation methods. The majority of the studies used either a mechanical or a kinematic definition, mostly focusing on the intrinsic behavior of the device and frequently giving little attention to the device impact of the user and on the user’s perception. Furthermore, user-centered or physiological assessments could be examined further, since evaluation metrics are usually based on kinematic and robot mechanical metrics. Only a few studies have examined the underlying motor control strategies, using more in-depth methods such as muscle synergy analysis. These findings highlight the need for a shared taxonomy and a standardized framework for transparency evaluation. Such efforts would enable more reliable comparisons between studies and support the development of more effective and user-centered READs. Full article