Search Results (348)

As unmanned aerial vehicles (UAVs) are increasingly deployed in mission-critical domains such as military operations, infrastructure inspection, and disaster response, the threat of GPS and network spoofing attacks has emerged as a fundamental challenge to operational continuity. Existing intrusion detection systems based on threshold rules or shallow machine learning models are inherently limited in their ability to identify the latent onset of sophisticated, gradually intensifying spoofing campaigns—a gap that motivates the present work. This study proposes a deep learning-based early detection and network resilience prediction framework that employs Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) architectures operating on three spatio-temporal network features—Hop Count Spike Rate (HCS), Packet Drop Volatility (PDV), and Spatial Disconnect Density (SDD)—proposed in this study. To reflect realistic adversarial conditions, we design a Gradual Adaptive Attacker model in which the spoofing intensity escalates progressively across six operational phases, including a second-stage adaptive attack that modulates its gradient upon detecting initial countermeasures. Both models are trained on 1000 simulated runs using sliding-window time-series sequences and evaluated across 500 independent test runs with paired statistical testing. The GRU model achieves a mean ROC-AUC of 0.9915 (±0.0091) and a mean F1-Score of 0.9099 (±0.0462), outperforming LSTM across all metrics with statistical significance at p < 0.001 under both the paired t-test and the Wilcoxon signed-rank test. Critically, GRU detects spoofing onset with an average latency of 0.503 time steps—approximately 29.4% faster than LSTM (0.712 steps)—a difference confirmed as statistically significant (p < 0.001, Cohen’s d = 0.41). Network resilience simulations further demonstrate that integrating GRU-based autonomous evasion maintains a Connectivity Ratio (CR) above 80% even under severe attack phases, whereas passive networks experience total connectivity collapse (CR = 0%). These findings establish GRU as the superior architecture for real-time UAV edge deployment and affirm that the proposed pipeline extends beyond threat alerting to actively preserving mission continuity under adversarial spoofing conditions. Full article

This paper presents a comprehensive UAV testbed that establishes quantitative baselines for hardware vulnerability diagnosis and cyber–physical security validation by leveraging comparative flight logs from PX4-based Software-In-The-Loop (SITL) simulations and multiple real-world quadrotor missions. The testbed utilizes a unified data pipeline centered on the uORB message bus and ULog format, enabling the extraction of high-resolution telemetry, including raw Inertial Measurement Unit (IMU) data, state-estimation, and actuator control signals. Evaluated across varying environmental conditions, side-by-side time-series and statistical analyses reveal critical sim-to-real discrepancies in sensor fidelity, GPS interference, and onboard resource behavior that are often overlooked in virtual environments. Real-world data exposes hardware-induced noise, mechanical vibrations, and electromagnetic disturbances that significantly impact flight stability and system reliability. By mathematically quantifying these discrepancies (e.g., via variance and probability distribution shifts), the proposed testbed establishes a rigorous baseline for distinguishing natural physical variability from anomalous or adversarial behavior. Ultimately, this work provides a foundational framework for developing robust anomaly detection models and validating the cyber–physical security of autonomous UAV systems in safety-critical environments. Full article

Electroencephalogram (EEG) signals, as a direct measure of the brain’s cortical electrophysiological activity, can objectively capture emotion-induced neural changes. Phase space reconstruction is an effective method for processing nonlinear time series. It maps time series to a high-dimensional phase space, thereby better preserving subtle dynamic information in the signal. This paper proposes a method for emotion recognition in EEG signals based on phase space reconstruction. First, the macro-topological features of the trajectories are constructed via phase space reconstruction. The time delay and embedding dimension are then optimized using the minimum cross-prediction error and the G-P method, followed by dimensionality reduction to a two-dimensional plane via local linear embedding. Building on this foundation, and in response to the limitations of manually designed features, we further propose a deep learning-based method for extracting multiscale dynamic features from trajectory images. The designed GN-MVXXS framework, which utilizes a granularity-adaptive module to adaptively switch the receptive field and a noise-filtering module to suppress isolated noise points, thereby effectively uncovers microscopic evolutionary features at the image level. Finally, to leverage the complementary strengths of macro- and micro-level information, we propose a fusion method based on dynamic attention. This approach aligns the dual representational dimensions through global average pooling and nonlinear dimension expansion, and utilizes a dynamic attention mechanism to adaptively assign feature weights, enabling the model to collaboratively enhance both overall dynamic patterns and local details based on sample characteristics. The experimental results show that the model achieved an accuracy of 96.11% in the three-class classification task on the SEED, 86.33% in the four-class classification task on the HIED, and 83.67% in classification across normal-hearing and hearing-impaired individuals, significantly outperforming single-feature models and traditional fusion methods. Full article

Accurate residual demand forecasting (RDF) is essential for stable peer-to-peer energy trading in developing economies. This study benchmarks three hyperparameter optimization paradigms, HEBO, PSO, and GP-BO, applied to XGBoost (2.1.4) forecasting using seven-fold TimeSeriesSplit validation on South African hourly grid data. Results demonstrate a fundamental trade-off between accuracy and efficiency: PSO achieves superior accuracy (0.47% MAPE) at the cost of substantial computation (23.4 h), while GP-BO offers revolutionary speed (19 min) with acceptable accuracy trade-offs. HEBO provides balanced performance with stable convergence. Crucially, we identify a “data–optimizer coupling” effect where optimal scaling methods are algorithm-dependent. These findings provide context-specific deployment strategies for microgrid operators addressing energy trilemma challenges. Full article

Existing studies of street-related emotional perception mainly rely on static scene evaluations, which cannot capture the cumulative effects of environmental exposure during continuous walking. To address this limitation, this study proposes a method for predicting pedestrian physiological responses in sequential audiovisual street environments. Four real-world walking routes were selected, with outbound and return directions treated as independent paths, yielding eight paths and 32 valid samples. EEG, ECG, sound pressure level, first-person video, and GPS data were synchronously collected to construct a 1 s multimodal time-series dataset. Pearson correlation, Kendall correlation, and mutual information analyses were used to examine linear, monotonic, and nonlinear relationships between environmental variables and physiological indicators, and the resulting weights were incorporated into a Long Short-Term Memory (LSTM) model for multi-step prediction. Visual elements and noise exposure were the main factors influencing physiological responses. Among the models, the mutual-information-weighted LSTM performed best, achieving an R² of 0.77 for heart rate variability (RMSSD), whereas prediction of the EEG ratio (β/α and θ/β) remained limited. An additional independent street sample outside the training set was then used to generate a dual-dimensional EEG-ECG physiological response map, demonstrating the model’s potential for identifying emotional risk segments and supporting street-level micro-renewal. Full article

Gas concentration series in coal mining faces are jointly affected by multiple coupled factors, including geological conditions, mining disturbances, ventilation organization, and gas drainage intensity, and therefore exhibit pronounced nonstationarity, strong fluctuations, spatiotemporal correlations across multiple monitoring points, and occasional abrupt spikes. To address these challenges, this study proposes a gas concentration prediction and early-warning method that integrates CEEMDAN–SST with GraphPINN-TimesFM (Graph Physics-Informed Neural Network–Time Series Foundation Model). First, based on multi-source monitoring data such as wind speed, gas concentrations at multiple monitoring points, and equipment operating status, anomaly removal, operating-condition segmentation, and change-point detection are performed to construct stable operating-state labels. Feature selection is then conducted by combining optimal time-lag correlation, Shapley value contribution, and dynamic time warping. Second, WGAN-GP is employed to augment samples from minority operating conditions, while CEEMDAN–SST is used to decompose and reconstruct the target series so as to reduce the interference of nonstationary noise and enhance sequence predictability. On this basis, TimesFM is adopted as the backbone for long-sequence forecasting to capture long-term dependency features in gas concentration evolution. Furthermore, GraphPINN is introduced to embed the topological associations among monitoring points, airflow transmission delays, and convection–diffusion mechanisms into the training process, thereby enabling collaborative modeling that integrates data-driven learning with physical constraints. Finally, the predictive performance, early-warning capability, and interpretability of the proposed model are systematically evaluated through regression forecasting, warning discrimination, and Shapley-based interpretability analysis. The results demonstrate that the proposed method can effectively improve the accuracy, robustness, and physical consistency of gas concentration prediction under complex operating conditions, thereby providing a new technical pathway for gas over-limit early warning and safety regulation in coal mining faces. Full article

Internet-of-Things (IoT) sensing can improve traceability, safety, and efficiency of medical waste handling, yet many deployments remain fragmented, lack an end-to-end system architecture, and do not provide the structured data pipelines needed for artificial intelligence (AI) analytics. This paper presents a layered IoT-based system design for medical waste management that integrates: (i) Espressif Systems 32 (ESP32)-based edge devices for fill-level and Global Positioning System (GPS) telemetry; (ii) secure network communication; (iii) a cloud backend for data ingestion, storage, and analytics; and (iv) operator dashboards with event-driven alerting. The architecture extends our prior GPS-enabled tracking and route optimization by adding sensor-driven state monitoring, threshold-based decision support, and a time-series data pipeline designed for future AI-driven predictive analytics. In a 30-day pilot with five containers, the system collected one reading every 15 min (14,400 total readings). The backend demonstrated efficient processing with an average Application Programming Interface (API) response time of 45 ms, sub-50 ms database write latency, and high uptime; alerts were delivered promptly upon threshold violations. Compared with a fixed-schedule baseline, the system enabled condition-based collection scheduling with zero data loss. The proposed design emphasizes modularity, fault tolerance, and integration readiness for hospital information systems, providing a practical blueprint for scalable smart-healthcare waste logistics and a foundation for machine learning-based predictive waste management. Full article

Global Positioning System (GPS)-derived ionospheric electron concentration measurements provide a powerful observational framework for seismo-electromagnetic studies, enabling quantitative investigation of lithosphere–atmosphere–ionosphere coupling processes through statistically detectable perturbations in ionospheric electron concentration. We analyze GPS-derived Vertical Total Electron Content (VTEC) variations associated with the 14 August 2021 Haiti earthquake ($M_w$ 7.2) and the 11 January 2022 Cyprus earthquake ($M_w$ 6.6) using data from nearby International GNSS (Global Navigation Satellite System) Service (IGS) stations located within their respective earthquake preparation zones. VTEC time series spanning 45 days before and 7 days after each event are processed to remove the diurnal component, yielding residuals that isolate short-term ionospheric variability. Anomaly detection is performed using three statistical frameworks: a Gaussian mean, standard deviation model, a robust median/median absolute deviation (MAD) model, and a distribution-free quantile-based model. Daily “occurrence” and “energy” indices are constructed to quantify the frequency and cumulative strength of detected anomalies, respectively. While the indices exhibit similar temporal patterns across all methods, they indicate frequent anomaly detection, limiting statistical selectivity. To address this, both indices are normalized by their median values and filtered using a 95% quantile threshold, retaining only extreme deviations. This procedure substantially reduces background fluctuations and isolates a small number of statistically significant anomaly peaks. For both earthquakes, enhanced anomaly activity is identified in the weeks preceding the events, whereas post-event peaks coincide with periods of elevated meteorological and geomagnetic activity. The results demonstrate that normalization combined with robust statistical methods is essential for discriminating significant ionospheric TEC anomalies from background variability. Full article

Here, we develop a digital community management application (APP) and an energy prediction and analysis system for smart communities. The system integrates the internet of things (IoT) technology and multiple prediction models to improve the intelligence and automation of community energy management. The developed APP has the following functions: user classification, announcement notification, express delivery management, GPS positioning navigation, calendar, and energy forecast. The hardware architecture of the system consists of a voltage/current sensing module, a Wireless Fidelity (Wi-Fi) module, and an Arduino platform, allowing real-time feedback and display of power consumption data. The energy forecasting part proposes a two-layer hybrid model architecture. This architecture combines Seasonal Trend decomposition using Loess (STL) time series decomposition, extreme gradient boosting (XGBoost), and Seasonal Autoregressive Integrated Moving Average (SARIMA) models to predict residential electricity consumption trends over the next 3 years. The results of the model prediction are verified using the data on Taiwan’s electricity consumption. The model accurately predicts the average monthly residential electricity consumption with a relative error of 5.8%, an acceptable energy management accuracy. This system integrates APP applications and efficient prediction models, demonstrating its great potential in smart community energy management and enhanced resident interaction. Full article

Crop yield estimation, particularly early-season yield prediction, is highly important for global food security and disaster mitigation. In this study, we utilized deep learning models combined with remote sensing data to develop in-season crop yield estimation models, enabling immediate yield prediction. We employed a convolutional neural network (CNN) for spatial feature extraction and a long short-term memory network (LSTM) for temporal patterns, complemented by Gaussian process regression (GP) that introduced geographical coordinates. Three groups of in-season yield prediction experiments were designed, utilizing four-phase, two-phase, and single-phase data, respectively. The results indicated that under the two-phase training scheme, the LSTM_GP model achieved the highest performance in the sixth period, with an $R^2$ value of 0.61 and a root mean square error ($RMSE$) value of 983.38 kg/ha. When trained on single-phase data at the twelfth phase (approximately mid-to-late July), the LSTM_GP model also performed best, attaining an $R^2$ value of 0.62 and an $RMSE$ value of 969.06 kg/ha. The single-phase prediction model outperformed time-series models in yield prediction accuracy. The periods from mid-to-late July to early-to-mid August represent critical crop growth stages were essential for accurate yield prediction. From our research, we found that adding GP can improve the prediction accuracy, especially for LSTM. Moreover, the proposed single-phase prediction model realized reliable crop yield prediction as well as the silking to early grain-filling stage (mid-to-late July), providing a critical lead time of approximately 2–2.5 months before harvest to support pre-harvest agricultural decision-making. Full article

With the rapid expansion of pulsar survey data driven by advanced radio telescopes such as FAST, automated detection methods have become crucial for the efficient and accurate identification of single-pulse signals. A key challenge in this task is the extreme class imbalance between genuine pulsar pulses and radio frequency interference (RFI), which significantly hampers classifier performance—particularly in low signal-to-noise ratio (S/N) environments. To address this issue and improve detection accuracy, we propose Pulsar-WRecon, a Wasserstein GAN with Gradient Penalty (WGAN-GP)-based framework designed to generate realistic single-pulse profiles. The synthetic samples generated by Pulsar-WRecon are used to augment training data and alleviate class imbalance. Building upon the enhanced dataset, Convolutional Kolmogorov–Arnold Network (CKAN) is further introduced as a novel hybrid model that integrates convolutional layers with KAN-based functional decomposition to better capture complex patterns in pulse signals. On the three-channel pulsar images from the HTRU1 dataset, our method achieves a recall of 97.5% and a precision of 98.5%. On the DM time series image dataset, FAST-DATASET, it achieves a recall of 93.2% and a precision of 92.5%. These results validate that combining generative data augmentation with an improved model architecture can effectively enhance the precision of single-pulse detection in large-scale pulsar surveys, especially in challenging, real-world conditions. Full article

Complications are frequent in cancer patients and contribute to adverse outcomes and higher healthcare costs, underscoring the need for earlier identification and prediction. This study evaluated the feasibility of using passively generated smartphone sensor data to explore early-warning signals of complications and symptom worsening during cancer treatment. A total of 108 patients were continuously monitored using accelerometer, GPS, and screen on/off data collected through the LAIMA application, while symptoms of depression, fatigue, and nausea were assessed every two weeks and complications were confirmed during clinic visits or emergency presentations. Smartphone data streams were aggregated into variables describing activity and sociability patterns. Machine learning models, including Decision Tree, Extreme Gradient Boosting, K-Nearest Neighbors, and Support Vector Machine, were used for complication prediction, and time-series models such as Autoregressive Integrated Moving Average, Holt–Winters, TBATS, Long Short-Term Memory neural network, and General Regression Neural Network were applied to identify early behavioral changes preceding symptom reports. In this exploratory analysis, the ensemble model demonstrated high sensitivity (89%) for identifying complication events. Smartphone-derived behavioral indicators enabled earlier detection of depression, fatigue, and vomiting by about nine days in a subset of patients. These findings demonstrate the feasibility of passive smartphone sensor data as exploratory early-warning signals, warranting validation in larger cohorts. Full article

Frequent cloud cover severely limits the use of optical remote sensing for continuous ecological monitoring. Synthetic aperture radar (SAR) offers an all-weather alternative, but translating SAR data to optical equivalents is challenging, particularly in cloudy regions where paired training data are scarce. To address this, we developed an enhanced CycleGAN (denoted by SA-CycleGAN) to derive a high-fidelity, temporally continuous normalized difference vegetation index (NDVI) from SAR imagery. The SA-CycleGAN introduces a novel spatiotemporal attention generator that dynamically computes global and local feature relationships to capture long-range spatial dependencies across diverse landscapes. Furthermore, a structural similarity (SSIM) loss function is integrated into the SA-CycleGAN to preserve the structural and textural integrity of the synthesized images. The performance of the SA-CycleGAN and three unsupervised models (DualGAN, GP-UNIT, and DCLGAN) was evaluated by deriving NDVI time series from Sentinel-1 SAR images across four sites with different vegetation types. Ablation experiments were conducted to verify the contributions of the key components in the SA-CycleGAN model. The results demonstrate that the SA-CycleGAN significantly outperformed the comparison models across all four sites. Quantitatively, the proposed method achieved the lowest Root Mean Square Error (RMSE) of 0.0502 and the highest Coefficient of Determination (R²) of 0.88 at the Zhangbei and Xishuangbanna sites, respectively. The ablation experiments confirmed that the attention mechanism and SSIM loss function were crucial for capturing long-range features and maintaining spatial structure. The SA-CycleGAN proves to be a robust and effective solution for overcoming data gaps in optical time series. Full article

Effective situation awareness relies on the robust processing of high-dimensional data streams generated by onboard sensors. However, the application of deep generative models to extract features from complex UAV sensor data (e.g., GPS, IMU, and radar feeds) faces two fundamental challenges: critical training instability and the difficulty of representing multi-modal distributions inherent in dynamic flight maneuvers. To address this, this paper proposes a novel unsupervised sensor data processing framework to overcome these issues. Our core innovation is a deep generative model, VAE-WRBM-MDN, specifically engineered for stable feature extraction from non-linear time-series sensor data. We demonstrate that while standard Variational Autoencoders (VAEs) often struggle to converge on this task, our introduction of Weighted-uncertainty Restricted Boltzmann Machines (WRBM) for layer-wise pre-training ensures stable learning. Furthermore, the integration of a Mixture Density Network (MDN) enables the decoder to accurately reconstruct the complex, multi-modal conditional distributions of sensor readings. Comparative experiments validate our approach, achieving 95.69% classification accuracy in identifying situational patterns. The results confirm that our framework provides robust enabling technology for real-time intelligent sensing and raw data interpretation in autonomous systems. Full article

Precipitable water vapor (PWV) is an important component of atmospheric spatial parameters and plays a vital role in meteorological studies. In this study, PWV retrieval by real-time precise point positioning (PPP) technique is validated by using global navigation satellite system (GNSS) observations and four real-time products from different analysis centers, which are Centre National d’Etudes Spatiales (CNES), Internation GNSS Service (IGS), Japan Aerospace Exploration Agency (JAXA), and Wuhan University (WHU). To comparatively analyze the performance of each scenario, the single-system (GPS/Galileo/BDS3), and multi-system (GPS + Galileo + BDS) PPP techniques are applied for zenith tropospheric delay (ZTD) and PWV retrieval. Then, the ZTD and PWV are evaluated by comparison with the IGS final ZTD product, the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) data, and radiosondes observations provided by the University of Wyoming. Experimental results demonstrate that the root mean squares error (RMS) of ZTD differences from multi-system solutions are below 11 mm with respect to the four-product series and the RMS of PWV differences are below 3.5 mm. As for single-system solution, the IGS real-time products lead to the worst accuracy compared with the other products. Besides the scenario of BDS3 observations with IGS real-time products, the RMS of ZTD differences from the GPS-only and Galileo-only solutions are all less than 15 mm compared to the four-product series, as well as the RMS of PWV differences is under 5 mm, which meets the accuracy requirement for GNSS atmosphere sounding. Full article