Vehicular Ad Hoc Networks (VANETs) are essential for secure and efficient communication at Intelligent Transportation Systems (ITS). But, in comparison, the dynamic and resource-constrained nature of VANETs poses design challenges to secure the network robustly while adhering to restrictions on computational and energy costs. This work presents a new ECC-based cryptographic scheme to prevent side-channel attacks (SCAs) and enhance speed in real-time vehicular operations. The scheme utilizes various countermeasures, including dynamic noise injection, masking, and optimized elliptic curve operations, to counteract main vulnerabilities and to optimize system performance. We get a detailed analysis of the proposed scheme concerning three prominent metrics, namely computational costs, communication overhead, and energy consumption, compared with existing works. The experiments showed notable enhancements, thus confirming a 29% increase in computational cost for mutual authentication, 25% for communication overhead for message signing and verification, and 26% for energy consumption in the given phases. These optimizations enable real-time performance, scalability, and energy efficiency in resource-constrained environments, ensuring the scheme’s suitability for deployment in VANETs. Moreover, the proposed scheme demonstrates robust security against several types of attacks, such as replay, impersonation, man-in-the-middle, and advanced side-channel attacks, thereby facilitating safe and reliable vehicular communications. The findings reported in this research lay the groundwork for a secure and efficient VANET, providing a pathway towards adaptive security mechanisms and real-world testing as future directions. Overall, this work presents a holistic perspective to tackle the issues of performance and security in future vehicular networks.
Hybrid aerial–aquatic ecosystems integrating Unmanned Aerial Vehicles (UAVs) with underwater drones are increasingly critical for maritime monitoring, coastal security, and environmental intelligence. However, reliable anomaly detection in such systems remains challenging due to dynamic operating conditions, non-independent and identically distributed (non-IID) multimodal data, accessibility constraints for operators with disabilities, and strict communication and privacy limitations. These factors significantly restrict the applicability of conventional centralized and unimodal approaches in real-world deployments. To address these challenges, this study proposes SwinGTANet, a federated and explainable anomaly detection framework tailored for hybrid UAV–underwater networks. The framework incorporates Multi-Scale Adaptive Signal Decomposition (MASD) for noise-resilient preprocessing and Statistical–Spectral–Graph Embedding with Entropy-Guided Balancing (SSGE-EGB) for compact and discriminative feature representation. The core model integrates Swin-based temporal attention, graph convolutional reasoning, and temporal convolutional modeling, while a neurosymbolic fusion layer enhances interpretability and operational trust. Extensive experiments on a real-world multimodal dataset demonstrate that SwinGTANet achieves 98.6% accuracy, 99.1% AUC, and 96.7% temporal consistency, while maintaining robustness under severe channel impairments, data sparsity, and federated non-IID conditions. With an inference latency of 1.56 ms per sample, the framework ensures real-time performance with reduced communication overhead. These results highlight the practical significance of SwinGTANet in enabling scalable, privacy-preserving, and interpretable anomaly detection, supporting reliable deployment of intelligent UAV–underwater systems in safety-critical maritime environments.
Fire hazards pose substantial risks on construction sites, where rapidly changing layouts, temporary installations, and obstructed circulation routes can hinder safe evacuation. This study evaluated a low-cost fire-safety training framework integrating Building Information Modeling and immersive Virtual Reality among 178 construction workers recruited from three active construction sites in Bangladesh. Fire-safety awareness was assessed using an eight-item instrument before training, immediately afterward, and at a two-week follow-up. Mean composite scores increased from 2.43 to 6.89, representing a mean improvement of 4.46 points and a very large paired-samples effect size of 3.79. Exact McNemar tests showed significant improvement across all eight fire-safety items. At follow-up, 162 participants achieved a mean score of 6.21, retaining 85.2% of the improvement above the pre-training baseline. Post-training simulated evacuation-route deviation averaged 6.59%. Multiple regression explained 33.8% of the variance in route deviation, with age and prior fire-safety training identified as significant predictors. These findings provide strong within-participant evidence of substantial and sustained improvement in fire-safety awareness following the low-cost immersive training framework, while demonstrating the feasibility of scalable assessment of simulated route adherence using mobile hardware.
This study investigates how holes and joints affect the stability of mining pillars through both experimental and numerical methods, using gypsum samples measuring 5 × 20 × 20 cm. Four distinct models were developed, each featuring varying numbers and configurations of holes and joints. In each model, a key joint was positioned at angles of 0°, 30°, 60°, 90°, 120°, and 150° relative to the horizontal plane, which also influenced the placement of the other joints and holes. The samples underwent uniaxial compression testing in conjunction with numerical modeling performed using PFC2D software. According to the results, new tensile cracks were first observed at the tip of the key joint oriented at 60°. The presence of holes affected crack growth, leading to increased branching as the number of holes increased. Notably, at the 60° angle, the decline in mechanical properties was more pronounced with the addition of holes and joints compared to other angles. Conversely, the model with a key joint angle of 90° demonstrated the highest mechanical properties. The fracture patterns predicted by the numerical models were consistently validated by the experimental observations, confirming the reliability and accuracy of the computational approach adopted in this study.
Accurate short-term electricity consumption forecasting for low-altitude economy enterprises remains challenging because of nonlinear dynamics, heterogeneous operating patterns, and noise-contaminated measurements. To address these issues, an efficient and robust forecasting framework, termed IMPA-RTBELM, is proposed by integrating an Improved Marine Predators Algorithm (IMPA) with a Regularized Twin-Boundary Extreme Learning Machine (RTBELM). Specifically, IMPA enhances the original Marine Predators Algorithm through Sobol sequence initialization to improve population diversity, dynamic opposition-based learning to alleviate premature stagnation, and pattern-search refinement to strengthen late-stage exploitation. RTBELM retains the closed-form training advantage of ELM while improving robustness via ridge-stabilized output learning and a twin-boundary clipping mechanism that suppresses impulsive prediction excursions. Benchmark-function experiments show that IMPA achieves faster convergence and better final solutions than the original MPA and several representative metaheuristics. Under four mixed-noise SinC regression settings, IMPA-RTBELM consistently improves generalization performance and substantially reduces forecasting errors, lowering the average test MAPE from 0.783% to 0.292% relative to RTBELM, while also reducing test MAE and RMSE by 12.98% and 12.50%, respectively. On real enterprise electricity consumption data, IMPA-RTBELM outperforms a broad range of statistical, machine-learning, and deep-learning baselines, achieving an RMSE of 127.0983, an MAE of 93.3518, and a MAPE of 0.78%. These results demonstrate that the proposed framework provides a favorable trade-off among accuracy, robustness, and efficiency for large-scale short-term electricity consumption forecasting.
The visual elements of a city play an important role in shaping the perceptual image of the city among its residents and visitors. However, with rapid urbanization in developing countries, issues have emerged that affect the formation of a clear image of the city, including transient and weak elements and a lack of integration among them. In Saudi Arabia, large cities such as Riyadh, Jeddah, and Dammam have recently witnessed rapid growth, with the likely consequences on their image. Despite the awareness of the importance of studying the image of cities among residents, it has been a neglected field for quite some time. Therefore, the purpose of this study is to explore how residents and visitors of large Saudi cities perceive the spatial elements and how important they consider the attributes of these elements towards forming a clear perceptual image. Data were collected using a questionnaire survey that was carried out in three Saudi cities: Riyadh, Jeddah, and Dammam. The statistical analysis, including hypothesis testing and ANOVA results show that a clear perception of the elements has the same trend as satisfaction with living in the cities under study. The results are validated by performing normality and homogeneity of the data. The clarity of the elements is found to be low in all the cities, but there are significant differences, and a strong pattern of such differences emerges as a conclusion. The research proves to be beneficial for the modification of the infrastructural planning and management flaws pointed out in the study and also it shall be helpful in town planning and design of new cities.
The rapid advances in cloud computing have led to an marked increase in the number of service providers, making it challenging for users to select the most appropriate vendor for their needs. Many service users prioritise trust and security when selecting the ideal service provider and ignore other equally important considerations, such as Quality of Service (QoS) and energy efficiency. This study introduces a cloud service selection and recommendation (CS-SR) using a Trust Computation Framework (TCF) and a lightweight blockchain mechanism that combines qualitative and quantitative QoS attributes. The lightweight lightweight blockchain in the proposed system addresses the privacy, security, scalability, and authentication challenges commonly encountered in many cloud platforms. Experimental comparison with existing selection processes the proposed technique achieves superior performance based on energy efficiency and execution time. It provides a effective solution that considers trust, security and user needs that lead to better cloud service selection.
Cobalt-based spinel nanostructures (MnCo2O4) were synthesized by a surfactant-free, ultrasound-assisted method using metal nitrate precursors. The influence of reaction parameters, including stoichiometric ratios, reaction time, and pH control via ammonia, was systematically investigated. The synthesized nanostructures were characterized using SEM,XRD,EDS,FT−IR,VSM, and DRD. XRD patterns confirmed the high crystalline purity of the material. The spectroscopic and microscopic methods proved the optical properties and the formation of the spinel phase, while microscopic and spectroscopic analyses verified the formation of homogeneous nanostructures, demonstrating the efficacy of the ultrasound-assisted synthesis approach. In addition to experimental work, the electronic structure of the MnCo2O4 spinel was modeled using spin-polarized GGA+U calculations. A set of Hubbard parameters from 0 to 8 was applied to the Mn and Co atoms which revealed the half-metallic behavior of the normal and inverse spinel for U>0, although when U=0, the normal spinel remains half-metal while the inverse spinel exhibits metallic behavior. Furthermore, time-dependent density functional perturbation theory was used to calculate the optical properties, and plasmonic peaks were identified at 1.59 and 3.56 eV and 1.40 and 3.93 eV for a transferred momentum q=0.1A°−1 along the [100] of normal and inverse spinel, respectively.
Automatic text classification is a central operation in digital libraries because cataloging pipelines must assign stable subject labels to records whose titles, abstracts, keywords, and subject fields are often incomplete, short, or mutually inconsistent. Strong transformer encoders improve text understanding, but prior library-oriented and generic document classifiers usually aggregate bibliographic metadata either by flattening fields into a single sequence or by using fusion modules that do not make missing-field uncertainty explicit. We propose DLTransClass, a transformer-based framework that represents each record as a heterogeneous metadata measure. The model encodes title, abstract, keyword, and subject-related fields with a shared transformer, fuses them through a Wasserstein metadata barycenter, and minimizes a missingness-aware Wasserstein objective whose first-order surrogate yields a local variation penalty on the fused representation. Under the Dirac-field construction used here, the barycenter reduces to a learned weighted mean, so the contribution is not a new closed-form transport solver by itself, but the coupling between field-aware fusion, record-specific uncertainty sets, and a stability-motivated training objective. We benchmark DLTransClass against 13 baselines, including a strong concatenated-field BERT baseline that isolates the benefit of barycentric fusion, alongside library-specific systems and 2024–2025 architectures such as Mamba-Doc and LongLoRA, and we report results on six datasets that span digital-library subject assignment and metadata-rich scholarly categorization. Experimental results show that DLTransClass achieves an average Macro-F1 of 82.2%, providing improved resilience under extreme metadata sparsity. In stress tests with 60% field removal driven by a realism-calibrated mask that mirrors empirically observed catalog incompleteness, our method retains an 8.1-point advantage over competitive fusion baselines while maintaining a production-ready inference latency of 9.8 ms per record on a single NVIDIA A100 GPU under fp16 inference with batch size 32.
Current network traffic anomaly detection systems fail to reliably identify rare zero-day exploits and privilege-escalation attacks because the extreme scarcity of minority-class samples prevents models from learning coherent decision boundaries for these categories. To close this gap, we propose PRISM (Prototype-guided Representation with Imbalanced Spatio-Temporal Modeling), a framework that simultaneously addresses dynamic graph topology and severe class imbalance. PRISM employs three key components. First, a Dynamic Adaptive Graph Construction (DAGC) mechanism that learns a task-sensitive sparse adjacency matrix end-to-end. Second, a Hierarchical Spatio-Temporal Encoder (HSTE) that fuses graph topology and causal flow history via bidirectional gating. Third, a prototype-conditioned Wasserstein GAN with Gradient Penalty (WGAN-GP) that anchors minority-class synthesis to dynamically updated semantic prototypes, directly addressing class imbalance without modifying the inference pipeline. The framework adds negligible inference overhead, requires no architectural modifications to the downstream classifier, and is therefore suitable for deployment in high-speed network monitoring environments. We evaluate PRISM on the NSL-KDD, CICIDS2017, and UNSW-NB15 public benchmarks. PRISM achieves peak F1-scores of 99.21%, 98.87%, and 93.47% on these datasets, representing improvements of 0.42%, 0.31%, and 0.73% over the baseline BiNF model.
Modular construction streamlines building processes by standardising elements or entire systems, enhancing efficiency and flexibility. Modular reinforced concrete solutions continue to pose challenges. This study addresses this gap by evaluating the real-scale performance of the MICADO-reinforced concrete system, which integrates frame-supported modular units with prefabricated panels installed simultaneously. Full-scale laboratory tests identified key challenges within the system, including compaction difficulties, inadequate concrete cover, and reinforcement congestion in splice and crimping zones. Despite these limitations, results confirm that PVC pipes can effectively function as permanent formwork for concrete columns, demonstrating the system's viability with necessary design optimizations. This study highlights the need for modifications in column cross-section dimensions and reinforcement layouts to improve constructability and compliance with standards. Future research should further explore the system's performance under varying loading conditions and column spacing to optimise its application in modular construction.
Stability and stabilization for finite-field (control) networks are investigated in this paper. For stability and stabilizability of finite-field (control) networks, necessary and sufficient algebraic conditions are obtained, respectively. Based on the decomposition of the state space, the open-loop controls are designed to stabilizing the system and the minimal time steps are estimated.
In this study, we present an artificial intelligence-based approach recurrent neural network (AI-RNN) to investigate bioconvective MHD flow of Eyring-Powell ternary nanofluids (EP-TNFM) under Radiative–Dissipative Effects. We use blood as a base fluid, which incorporates silver, copper, and aluminum nanoparticles. Two fluid arrangements were employed, specifically one fluid being a hybrid nanofluid (Ag+Cu+blood) and the second, a ternary hybrid nanofluid (Ag+Cu+Al+blood). The governing nonlinear ordinary differential equations were developed using similarity transformations and were solved using boundary value analysis approaches. The datasets composed of converged numerical solutions over a series of parametric values, generated from above numerical model, were used for training and validation of AI-RNN, and testing. The procedure presented strong ability to converge quickly, predictive reliability, and provided evidence that both the training, validation, and testing loss are practically identical, R² values at least 0.98 regression results in terms of the training loss. Gradient magnitude and weight distribution indicated that the models were stable. The parametric results showed that for skin friction, Le decreased and α increased, whereas for Nusselt number, Le is decreased, and Pr is increased. Velocity and thermal boundary layer were heavily influenced by α, M, λ, and Pr. The higher Sc and Le numbers suppressed mass and microorganism transport. In terms of comparisons, ternary nanofluids had smaller velocities and reduced bioconvection than hybrid nanofluids due to the larger viscosity and density. Thus, results demonstrate that AI-RNN is a novel and reliable alternative to traditional solvers for hybrid nanofluids, with potential applications in biomedical engineering, as well as thermal management and microfluidic applications. Further work could expand on this work in several ways, by including other losses such as MAE and RMSE, using different architectures like LSTMs and PINNs, and by investigating the application of the approach to more complex non-Newtonian models and actual biomedical flows.
The rapid proliferation of IoT has made resource-constrained devices prime targets for cyber intrusions, where severe sample space overlap in high-dimensional traffic severely hampers effective detection. We propose a novel manifold reconstruction framework that combines a bilinear-conditional variational autoencoder (B-CVAE) with adaptive batch-level graph Laplacian regularization. The B-CVAE captures high-order feature-condition interactions via bilinear mapping and residual fusion, while the Laplacian term dynamically preserves local manifold geometry to enhance intra-class compactness and inter-class separation. A lightweight classifier is applied to the reconstructed representations for final detection. Extensive experiments on NSL-KDD, RT-IoT2022, and CIC-IoT2023 datasets demonstrate mean accuracies reaching 99.6% with low standard deviations across multiple runs, outperforming recent autoencoder-based and hybrid research. The method offers a robust, efficient solution tailored to resource-limited IoT environments.
Wheeled-legged quadruped robots exhibit strong adaptability to complex terrain environments. However, reliable perception of special terrains such as stairs and slopes remains a critical challenge for autonomous motion decision-making. To address this problem, this study proposes a vision-based special terrain perception and distance estimation system for wheeled-legged quadruped robots. This system develops a wavelet-enhanced real-time detection transformer, known as WLRT-DETR, combined with binocular depth estimation, achieving precise recognition and distance estimation of special terrain targets. First, to achieve the lightweight of the feature extraction network, we design a Re-parameterized Cross-stage Partial Aggregation (ReCSPA) module. By leveraging a dual-path structure combined with channel compression and structural re-parameterization strategies, the module effectively reduces the number of model parameters. Second, to enhance edge representation in feature maps, we introduce a Wavelet-Spatial Fusion Enhancement (WSFE) module, which exploits discrete wavelet transform (DWT) to decompose the frequency domain information of images and deeply fuse it with spatial features, thereby improving detection accuracy. Finally, the candidate boxes generated by WLRT-DETR are used as prior information to constrain the disparity search space, within which the Semi-Global Block Matching (SGBM) algorithm is applied to estimate the spatial distance of the targets, providing reliable depth information for robotic motion decision-making. Extensive experiments were conducted on a self-constructed special terrain dataset to comprehensively evaluate the proposed model by comparing it with state-of-the-art detectors, including representative YOLO-based models and RT-DETR v1/v2. The experimental results demonstrate that the proposed WLRT-DETR achieves superior performance in terms of precision, F1-score, and mean average precision (mAP) while maintaining a lightweight architecture suitable for real-time deployment. Moreover, the stereo distance estimation achieves a relative depth error within 3%, satisfying the practical requirements of wheeled-legged quadruped robots operating in special terrains.
Camera-based small object detection is essential for intelligent traffic surveillance, yet existing YOLO variants, Transformer-enhanced detectors, and lightweight traffic detection models still struggle to jointly satisfy detection accuracy, real-time inference, and edge deployment requirements under complex working conditions. To address this problem, this paper proposes YOLOv11-TSO, a Transformer-enhanced lightweight detector specifically designed for traffic small object detection in adverse environments. The central objective is to improve the detection of distant pedestrians, compact vehicles, bicycles, and partially occluded targets while maintaining practical inference efficiency on resource-constrained devices. YOLOv11-TSO introduces three coordinated innovations: a C2PSA-SA backbone that embeds Transformer self-attention for lightweight global contextual modeling, a C3k2_GAM neck that integrates group attention and refined multi-scale feature fusion, and label smoothing loss to mitigate overfitting and class imbalance. Extensive experiments on VisDrone2019 and WiderPerson, together with robustness tests under fog, rain, occlusion, and low-light conditions, demonstrate the effectiveness of the proposed model. On VisDrone2019, YOLOv11-TSO achieves 93.7% precision, 96.1% recall, and 92.4% mAP@0.5 with only 9.2M parameters and 34.0 GFLOPs. It reaches 35 FPS on NVIDIA H100, 18 FPS on Jetson TX2, and 8 FPS on Jetson Nano, indicating a practical accuracy–efficiency–deployment trade-off for intelligent traffic surveillance. These results confirm that YOLOv11-TSO provides a robust and lightweight solution for small object detection in complex traffic environments.
