Search Results (268)

Federated learning (FL) is a promising privacy-preserving paradigm for machine learning in distributed environments. Although FL reduces communication overhead, it does not itself provide low-latency guarantees. In IIoT environments, real-time responsiveness is primarily enabled by edge computing and local inference, while FL contributes indirectly by minimizing the need to transmit raw data across the network. This paper explores the use of FL for intrusion detection in IIoT networks and compares its performance with traditional centralized machine learning approaches. A simulated IIoT environment was developed in which each node locally trains a model on synthetic normal and attack traffic data, sharing only model parameters with a central server. The Flower framework was employed to coordinate training and model aggregation across multiple clients without exposing raw data. Experimental results show that FL achieves detection accuracy comparable to centralized models while significantly reducing privacy risks and network transmission overhead. These results demonstrate the feasibility of FL as a secure and scalable solution for IIoT intrusion detection. Future work will validate the approach on real-world datasets and heterogeneous edge devices to further assess its robustness and effectiveness. Full article

The rapid development of the Internet of Things (IoT), a network of interconnected devices and sensors, has improved operational efficiency, comfort, and sustainability in smart buildings. However, relying on interconnected systems also introduces cybersecurity vulnerabilities. For instance, attackers can exploit zero-day vulnerabilities (previously unknown security flaws), launch Distributed Denial of Service (DDoS) attacks (overwhelming network resources with traffic), or access sensitive Building Management Systems (BMS, centralized platforms for controlling building operations). By targeting critical assets such as Heating, Ventilation, and Air Conditioning (HVAC) systems, security cameras, and access control networks, they may compromise the safety and functionality of the entire building. To address these threats, this paper presents a cybersecure intelligent sensor framework to protect smart buildings from various IoT-related cyberattacks. The main component is an automated Intrusion Detection System (IDS, software that monitors network activity for suspicious actions), which uses machine learning algorithms to rapidly identify, classify, and respond to potential threats. Furthermore, the framework integrates intelligent sensor networks with AI-based analytics, enabling continuous monitoring of environmental and system data for behaviors that might indicate security breaches. By using predictive modeling (forecasting attacks based on prior data) and automated responses, the proposed system enhances resilience against attacks such as denial of service, unauthorized access, and data manipulation. Simulation and testing results show high detection rates, low false alarm frequencies, and fast response times, thereby supporting the cybersecurity of smart building infrastructures and minimizing downtime. Overall, the findings suggest that AI-enhanced cybersecurity systems offer promise for IoT-based smart building security. Full article

To identify the main controlling factors of well shutdowns in different production stages of coalbed methane (CBM) horizontal wells, this study investigated the production parameters and pump inspection records of 25 horizontal wells in Huabei Oilfield. This paper first summarizes the types, causes, and impact degrees of well shutdown faults. Then, it conducts an analysis focusing on the four core production stages—water drainage, production increase, stable production, and production reduction—and clarifies that the key fault difference across stages lies in the variation in main fault types. The following results show that: (1) a total of 15 types of shutdown faults occurred during production, which are classified into four categories: coal–sand mixture-related faults, gas intrusion-related faults, supporting equipment faults, and other faults. Coal–sand mixture are the core inducement (accounting for 52%), followed by gas intrusion (accounting for 22%). (2) The impact of faults varies significantly: wellbore blockage, pump sticking, and flexible shaft breakage caused by coal–sand mixture and high current due to gas intrusion have a significant impact on production; environmental protection issues only occur in the water drainage stage and do not affect production; supporting equipment faults have a short handling cycle and minimal impact. (3) Shutdown faults exhibit obvious stage characteristics: In the water drainage stage, faults are mainly caused by environmental protection, power outage, and other factors, while high current due to pump sticking is the core downhole fault; in the production increase and stable production stages, pump sticking and flexible shaft breakage induced by coal–sand mixture are dominant; in the production decline stage, gas intrusion problems intensify, and the proportion of coal–sand mixture -related faults decreases but remains the main inducement. Full article

The proliferation of Internet of Medical Things (IoMT) devices in healthcare requires robust intrusion detection systems to protect sensitive data and ensure patient safety. While existing neural network-based Intrusion Detection Systems have shown considerable effectiveness, significant challenges persist—particularly class imbalance and high data dimensionality. Although various approaches have been proposed to mitigate these issues, their actual impact on detection accuracy remains insufficiently explored. This study investigates advanced Artificial Neural Network (ANN) architectures and preprocessing strategies for intrusion detection in IoMT environments, addressing critical challenges of feature dimensionality and class imbalance. Leveraging the WUSTL-EHMS-2020 dataset—a specialized dataset specifically designed for IoMT cybersecurity research—this research systematically examines the performance of multiple neural network designs. Our research implements and evaluates five distinct ANN architectures: the Standard Feedforward Network, the Enhanced Channel ANN, Dual-Branch Addition and Concatenation ANNs, and the Shortcut Connection ANN. To mitigate the class imbalance challenge, we compare three balancing approaches: the Synthetic Minority Over-sampling Technique (SMOTE), Hybrid Over-Under Sampling, and the Weighted Cross-Entropy Loss Function. Performance analysis reveals nuanced insights across different architectures and balancing strategies. SMOTE-based models achieved average AUC scores ranging from 0.8491 to 0.8766. Hybrid sampling strategies improved performance, with AUC increasing to 0.8750. The weighted cross-entropy loss function demonstrated the most consistent performance. The most significant finding emerges from the Dual-Branch ANN with addition operations and a weighted loss function, which achieved 0.9403 Accuracy, 0.8786 AUC, a 0.8716 F1-Score, 0.8650 Precision, and 0.8786 Recall. Compared to the related work’s baseline, it demonstrates a substantial increase in F1 Score by 8.45% and an improvement of 18.67% in AUC and Recall, highlighting the model’s superiority at identifying potential security threats and minimizing false negatives. Full article

The rapid expansion of the Industrial Internet of Things (IIoT) within smart grid infrastructures has increased the risk of sophisticated cyberattacks, where severe class imbalance and stringent real-time requirements continue to hinder the effectiveness of conventional intrusion detection systems (IDSs). Existing approaches often achieve high accuracy on specific datasets but lack generalizability, interpretability, and stability when deployed across heterogeneous IIoT environments. This paper introduces AegisGuard, a hybrid intrusion detection framework that integrates an adaptive four-stage sampling process with a calibrated ensemble learning strategy. The sampling module dynamically combines SMOTE, SMOTE-ENN, ADASYN, and controlled under sampling to mitigate the extreme imbalance between benign and malicious traffic. A quantum-inspired feature selection mechanism then fuses statistical, informational, and model-based significance measures through a trust-aware weighting scheme to retain only the most discriminative attributes. The optimized ensemble, comprising Random Forest, Extra Trees, LightGBM, XGBoost, and CatBoost, undergoes Optuna-based hyperparameter tuning and post-training probability calibration to minimize false alarms while preserving accuracy. Experimental evaluation on four benchmark datasets demonstrates the robustness and scalability of AegisGuard. On the CIC-IoT 2023 dataset, it achieves 99.6% accuracy and a false alarm rate of 0.31%, while maintaining comparable performance on TON-IoT (98.3%), UNSW-NB15 (98.4%), and Bot-IoT (99.4%). The proposed framework reduces feature dimensionality by 54% and memory usage by 65%, enabling near real-time inference (0.42 s per sample) suitable for operational IIoT environments. Full article

The integration of federated learning into Industrial Internet of Things (IIoT) networks enables collaborative intelligence but also exposes systems to identity spoofing, model poisoning, and malicious update injection. This paper presents Leash-FL, a lightweight self-healing framework that combines certificateless elliptic curve cryptography with blockchain to enhance resilience in resource-constrained IoT environments. Certificateless ECC with pseudonym rotation enables efficient millisecond-scale authentication with minimal metadata, supporting secure and unlinkable participation. A similarity-governed screening mechanism filters poisoned and free-rider updates, while blockchain-backed checkpoint rollback ensures rapid recovery without service interruption. Experiments on intrusion detection, anomaly detection, and vision datasets show that Leash-FL sustains over 85 percent accuracy with 50 percent malicious clients, reduces backdoor success rates to under 5 percent within four recovery rounds, and restores accuracy up to three times faster than anomaly-screening baselines. The blockchain layer achieves low-latency consensus, high throughput, and modest ledger growth, significantly outperforming Ethereum-based systems. Membership changes are efficiently managed with sub-50 ms join and leave operations and re-admission within 60 ms, while guaranteeing forward and backward secrecy. Leash-FL delivers a cryptography-driven approach that unifies lightweight authentication, blockchain auditability, and self-healing recovery into a secure, resilient, and scalable federated learning solution for next-generation IIoT networks. Full article

Major depressive disorder (MDD) affects approximately $4.4\%$ of the global population. Its prevalence is increasing among adolescents and has led to the psychosocial condition known as hikikomori. MDD is typically assessed by self-report questionnaires, which, although informative, are subject to evaluator bias and subjectivity. To address these limitations, recent studies have explored machine learning (ML) for automated MDD detection. Among the input data used, speech signals stand out due to their low cost and minimal intrusiveness. However, many speech-based approaches lack integration with cognitive behavioral therapy (CBT) and adherence to evidence-based, patient-centered care—often aiming to replace rather than support clinical monitoring. In this context, we propose ML models to assess MDD in hikikomori patients using speech data from a real-world clinical trial. The trial is conducted in Italy, supervised by physicians, and comprises an eight-session CBT plan that is clinical evidence-based and follows patient-centered practices. Patients’ speech is recorded during therapy, and the Mel-Frequency Cepstral Coefficients (MFCCs) and wav2vec 2.0 embedding are extracted to train the models. The results show that the Multi-Layer Perceptron (MLP) predicted depression outcomes with a Root Mean Squared Error (RMSE) of $0.064$ using only MFCCs from the first session, suggesting that early-session speech may be valuable for outcome prediction. When considering the entire CBT treatment (i.e., all sessions), the MLP achieved an RMSE of $0.063$ using MFCCs and a lower RMSE of $0.057$ with wav2vec 2.0, indicating approximately a $9.5\%$ performance improvement. To aid the interpretability of the treatment outcomes, a binary task was conducted, where Logistic Regression (LR) achieved $70\%$ recall in predicting depression improvement among young adults using wav2vec 2.0. These findings position speech as a valuable predictive tool in clinical informatics, potentially supporting clinicians in anticipating treatment response. Full article

We present two case studies of tropospheric aerosol transport observed over the high-altitude Helmos observatory (1800–2300 m a.s.l.) in Greece during September 2021. Two cases were linked to Saharan dust intrusions, of which one was additionally linked to a mixture of biomass-burning and continental aerosols. Aerosol vertical profiles from the AIAS mobile backscatter/depolarization lidar (532 nm, NTUA) revealed distinct aerosol layers between 2 and 6 km a.s.l., with particle linear depolarization ratio values of up to 0.30–0.40, indicative of mineral dust. The elevated location of Helmos allows lidar measurements in the free troposphere, minimizing planetary boundary layer influence and improving the attribution of long-range transported aerosols. Radiative impacts were quantified using the LibRadtran model. For the 27 September dust outbreak, simulations showed strong shortwave absorption within 3–7 km, peaking at 5–6 km, with surface forcing reaching −25 W m⁻² and TOA forcing around −12 W m⁻², thus, implying a net cooling by 13 W m⁻² on the Earth’s atmosphere system. In contrast, the 30 September mixed aerosol case produced substantial solar attenuation, a surface heating rate of 2.57 K day⁻¹, and a small positive forcing aloft (~0.05 K day⁻¹). These results emphasize the contrasting radiative roles of dust and smoke over the Mediterranean and the importance of high-altitude observatories for constraining aerosol–radiation interactions. Full article

This review explores the transformative potential of graphene-based, non-invasive biochemical sensors in the context of real-time health monitoring and personalised medicine. Traditional diagnostic methods often involve invasive procedures that can be uncomfortable, pose risks, and limit the frequency of monitoring. In contrast, wearable sensors incorporating graphene offer a compelling alternative by enabling continuous, real-time tracking of physiological and biochemical signals with minimal intrusion. Graphene’s exceptional electrical conductivity, mechanical flexibility, biocompatibility, and high surface-area-to-volume ratio make it ideally suited for integration into skin-conformal sensor platforms. These properties not only enhance sensitivity and signal fidelity but also promote user comfort and long-term wearability, critical factors for the adoption of wearable health technologies. The discussion evaluates current developments in the design and deployment of graphene-based biosensors, with particular attention given to their role in managing chronic conditions, supporting preventative healthcare, and facilitating decentralised diagnostics. By bridging materials science and biomedical engineering, this review positions graphene as a key enabler in the shift towards more proactive, patient-centred healthcare models. The text also identifies ongoing challenges and future directions in sensor design, aiming to inform researchers working at the intersection of advanced materials and medical technology. Full article

Intrusion Detection Systems (IDS) are vital to cybersecurity but suffer from severe class imbalance in benchmark datasets such as NSL-KDD and UNSW-NB15. Conventional oversampling methods (e.g., SMOTE, ADASYN) are efficient yet fail to preserve the latent semantics of rare attack behaviors. This study introduces the Minority-class Intrusion Detection Synthesizer GAN (MIDS-GAN), a divergence-minimization framework for minority data augmentation under structured feature constraints. MIDS-GAN integrates (i) correlation-based structured feature selection (SFS) to reduce redundancy, (ii) trainable ACON activations to enhance generator expressiveness, and (iii) KL-divergence-guided alignment to ensure distributional fidelity. Experiments on NSL-KDD and UNSW-NB15 demonstrate significant improvement on detection, with recall increasing from 2% to 27% for R2L and 1% to 17% for U2R in NSL-KDD, and from 18% to 44% for Worms and 69% to 75% for Shellcode in UNSW-NB15. Weighted F1-scores also improved to 78%, highlighting MIDS-GAN’s effectiveness in enhancing minority-class detection through a principled, divergence-aware approach. Full article

The widespread integration of Internet-connected devices into industrial environments has enhanced connectivity and automation but has also increased the exposure of industrial cyber–physical systems to security threats. Detecting anomalies is essential for ensuring operational continuity and safeguarding critical assets, yet the dynamic, real-time nature of such data poses challenges for developing effective defenses. This paper introduces DataSense, a comprehensive dataset designed to advance security research in industrial networked environments. DataSense contains synchronized sensor and network stream data, capturing interactions among diverse industrial sensors, commonly used connected devices, and network equipment, enabling vulnerability studies across heterogeneous industrial setups. The dataset was generated through the controlled execution of 50 realistic attacks spanning seven major categories: reconnaissance, denial of service, distributed denial of service, web exploitation, man-in-the-middle, brute force, and malware. This process produced a balanced mix of benign and malicious traffic that reflects real-world conditions. To enhance its utility, we introduce an original feature selection approach that identifies features most relevant to improving detection rates while minimizing resource usage. Comprehensive experiments with a broad spectrum of machine learning and deep learning models validate the dataset’s applicability, making DataSense a valuable resource for developing robust systems for detecting anomalies and preventing intrusions in real time within industrial environments. Full article

Geohazards pose significant dangers to human safety, infrastructures, and the environment, highlighting the need for advanced monitoring techniques for early damage detection and structure management. The distributed optical fiber sensors (DFOS) are strain, temperature, and vibration monitoring tools characterized by minimal intrusiveness, accuracy, ease of deployment, and the ability to perform measurements with high spatial resolution. Although these sensors rely on well-established measurement techniques, available for over 40 years, their diffusion within monitoring and early warning systems is still limited, and there is a certain mistrust towards them. In this regard, based on several case studies, the implementation of DFOS for early warning of various geotechnical hazards, such as landslides, earthquakes and subsidence, is discussed, providing a comparative analysis of the typical advantages and limitations of the different systems. The results show that real-time monitoring systems based on well-established distributed fiber-optic sensing techniques are now mature enough to enable reliable and long-term geotechnical applications, identifying a market segment that is only minimally saturated by using other monitoring techniques. More challenging remains the application of the technique for vibration detection that still requires improved interrogation technologies and standardized practices before it can be used in large-scale, real-time early warning systems. Full article

In the context of increasing urban night lighting, the phenomenon of light trespass in residential areas is becoming increasingly serious, affecting the night comfort and circadian rhythm of residents. Aiming at this problem, this paper takes the night lighting of activity places in old multi-story residential areas of Shijingshan, Beijing, as the research object, and proposes a research framework integrating parametric modeling, multi-objective optimization, correlation analysis, and scheme decision-making, aiming to trade off the two objectives of maximizing the night lighting of the activity places and minimizing indoor light intrusiveness. The study first establishes a parametric model based on Rhino and Grasshopper, combines the NSGA-II algorithm with multi-objective optimization simulation to obtain the Pareto optimal solution, analyzes the correlation between the design variables and the objective function by the Spearman method, and finally assists in the scheme decision-making by K-means clustering. The results showed that the streetlight heights (SH), distance between buildings and streetlights (DBS), and streetlight matrix types (SMT) were the key factors affecting lighting performance, which should be emphasized in the actual lighting design. Secondly, the Cluster2 solution set optimally performs the two objective functions. The 18th individual of Generation 15 (Gen. 15 Ind. 18) and Gen. 31 Ind. 42 are recommended, providing practical guidance for night lighting design in residential areas. The innovation of this study lies in applying multi-objective optimization and K-means clustering to optimize the night lighting environment in micro-spaces within old multi-story residential areas in cities, offering new insights for lighting design in similar scenarios. Full article

As network automation and self-organizing networks (SONs) rapidly evolve, edge devices increasingly demand lightweight, real-time, and high-precision classification algorithms to support critical tasks such as traffic identification, intrusion detection, and fault diagnosis. In recent years, cross-entropy (CE) loss has been widely adopted in deep learning classification tasks due to its computational efficiency and ease of optimization. However, traditional CE methods primarily focus on class separability without explicitly constraining intra-class compactness and inter-class boundaries in the feature space, thereby limiting their generalization performance on complex classification tasks. To address this issue, we propose a novel classification loss framework—Contrastive Geometric Cross-Entropy (CGCE). Without incurring additional computational or memory overhead, CGCE explicitly introduces learnable class representation vectors and constructs the loss function based on the dot-product similarity between features and these class representations, thus explicitly reinforcing geometric constraints in the feature space. This mechanism effectively enhances intra-class compactness and inter-class separability. Theoretical analysis further demonstrates that minimizing the CGCE loss naturally induces clear and measurable geometric class boundaries in the feature space, a desirable property absent from traditional CE methods. Furthermore, CGCE can seamlessly incorporate the prior knowledge of pretrained models, converging rapidly within only a few training epochs (for example, on the CIFAR-10 dataset using the ViT model, a single training epoch is sufficient to reach 99% of the final training accuracy.) Experimental results on both text and image classification tasks show that CGCE achieves accuracy improvements of up to 2% over traditional CE methods, exhibiting stronger generalization capabilities under challenging scenarios such as class imbalance, few-shot learning, and noisy labels. These findings indicate that CGCE has significant potential as a superior alternative to traditional CE methods. Full article

Industrial control systems (ICSs) face escalating security challenges due to evolving cyber threats and the inherent limitations of traditional intrusion detection methods, which fail to adequately model spatiotemporal dependencies or interpret complex protocol semantics. To address these gaps, this paper proposes DLG–IDS—a lightweight intrusion detection framework that innovatively integrates dynamic graph construction for capturing real–time device interactions and logical control relationships from traffic, LLM–driven semantic enhancement to extract fine–grained embeddings from graphs, and a spatio–temporal graph neural network (STGNN) optimized via sparse attention and local window Transformers to minimize computational overhead. Evaluations on SWaT and SBFF datasets demonstrate the framework’s superiority, achieving a state–of–the–art accuracy of 0.986 while reducing latency by 53.2% compared to baseline models. Ablation studies further validate the critical contributions of semantic fusion, sparse topology modeling, and localized temporal attention. The proposed solution establishes a robust, real–time detection mechanism tailored for resource–constrained industrial environments, effectively balancing high accuracy with operational efficiency. Full article