Search Results (5)

As hardware Trojans (HTs) are becoming increasingly stealthy in global semiconductor supply chains, the need for both robust and explainable detection methods is pressing. The use of deep learning models (e.g., Siamese networks, Transformer models) in side-channel signals has shown promising detection accuracy. Yet, they are black-box, data-intensive, and do not expose the causal, structural, and temporal relationships that indicate the presence of HTs. In this paper, we present a causality-focused and explainable detection framework that goes beyond pattern matching. We develop a Self-Supervised Spatio-Temporal Graph Neural Network (SST-GNN) that embeds spatio-temporal side-channel information. Our approach builds a graph that models gate-level components as nodes with temporal power and electromagnetic (EM) features, and functional and physical connections as edges. To address label scarcity, a common problem in real-world applications, we leverage a self-supervised pretraining approach. In particular, a context-aware contrastive loss allows the model to differentiate valid augmentations of benign subgraphs and their side-channel signatures, thus capturing general representations of benign components without Trojan labels. This involves a Causality-Aware GNN (CA-GNN) layer, which embeds differentiable causal discovery into graph learning. This process decouples correlation from causation, identifying the pathways potentially affected by HT trigger and payload. To explain decision making, a gradient-based graph explainer localizes minimal decisive subcircuits and pivotal time windows, generating intuitive detection reports. We evaluated our method on the IEEE Hardware Trojan Side-Channel Dataset (with netlist data), achieving state-of-the-art results (F1 > 0.98). In particular, the model achieves over 60% improvement in Trojan localization precision and false-positive rate, compared to Transformer-based approaches, with high label efficiency and adversarial robustness. Full article

As autonomous driving technology progresses, efficient and accurate object detectors are able to detect pedestrians, vehicles, road signs, and obstacles in real time, thereby enhancing driving safety and serving as a part of autonomous driving. However, the performance of such object detectors is limited and cannot be leveraged to satisfy modern autonomous driving systems. To address this issue, we develop an object detection network for autonomous driving scenarios, SST-YOLO, which is based on YOLOv8. First, we propose a Sobel Convolution & Convolution (SCC) module to enhance the backbone, which incorporates a SobelConv branch to explicitly model gradient-based edge information and improve structural feature representation. In addition, we replace the original path aggregation feature pyramid network (PAFPN) with a Small Object Augmentation Pyramid Network (SOAPN), which integrates SPDConv and CSP-OmniKernel modules to strengthen multi-scale feature fusion and enhance small object representation. Finally, a Task-Adaptive Decomposition & Alignment Head (TADAHead) is designed, which employs task decomposition, dynamic deformable convolution, and classification-aware modulation to decouple tasks and achieve adaptive spatial alignment, thereby improving detection accuracy and robustness in complex scenarios. Experiments on the public autonomous driving dataset KITTI show that our proposed method outperforms the baseline YOLOv8 model. Compared with the baseline results, mAP@0.5:0.95 ranges from 65.1% to 69.2%, which indicates that the proposed SST-YOLO network can achieve object detection for autonomous cars. Full article

The fish catch in natural upwelling areas, which accounts for only 0.1% of the ocean surface, accounts for more than 40% of the world’s catch. The Zhoushan fishery, which is the largest fishery in China, is mainly formed by a coastal upwelling that features a low temperature. The upwelling in the study area (29.5–31.5° N, 121.5–123.5° E) is a vital factor affecting the formation of the Zhoushan fishery, and the primary productivity and low temperature that are brought by the rising water are important features of the upwelling. This study used global real-time high-resolution multivariate fused satellite (OSTIA) daily sea surface temperature (SST) data developed by the United Kingdom Meteorological Office that were collected from 1981 to 2020 to explore the spatial and temporal variation of the characteristics of the upwelling phenomenon in the study area. The data were processed by a temperature gradient-based upwelling edge detection algorithm to extract information on the central location of the upwelling, the location clusters in the core area, and the intensity index. The quantities of center and core area clusters were counted for each pixel point, and their corresponding probability values were calculated. The results of the spatial and temporal variation of the characteristics of the upwelling show that the upwelling in the study area was generated in April of each year, increased in intensity, and peaked in August, furthermore, the southern part of the upwelling dissipated in September. The region’s upwelling is spatially oblique and elliptical, with its long axis following the northeast and extending as far as the mouth of the Yangtze River. Its central location and core area were relatively stably existing in Ma’an Archipelago and Zhongjieshan Islands, which was consistent with the location of the two marine pastures in Zhoushan. According to our findings, locations with higher probability values in the upwelling center and core area, where upwelling occurs frequently, are usually accompanied by higher productivity and offer the potential to develop fishing grounds. The insights that were drawn from the study observations can, therefore, provide some reference for future artificial upwelling site selection. Full article

The identification of oceanographic circulation related features is a valuable tool for environmental and fishery management authorities, commercial use and institutional research. Remote sensing techniques are suitable for detection, as in situ measurements are prohibitively costly, spatially sparse and infrequent. Still, these imagery applications require a certain level of technical and theoretical skill making them practically unreachable to the immediate beneficiaries. In this paper a new geospatial web service is proposed for providing daily data on mesoscale oceanic feature identification in the North Aegean Sea, produced by Sentinel-3 SLSTR Sea Surface Temperature (SST) imagery, to end users. The service encompasses an automated process for: raw data acquisition, interpolation, oceanic feature extraction and publishing through a webGIS application. Level-2 SST data are interpolated through a Co-Kriging algorithm, involving information from short term historical data, in order to retain as much information as possible. A modified gradient edge detection methodology is then applied to the interpolated products for the mesoscale feature extraction. The resulting datasets are served according to the Open Geospatial Consortium (OGC) standards and are available for visualization, processing and download though a dedicated web portal. Full article

The present study focuses on the impact of the underlying RANS turbulence model in the Zonal Detached Eddy Simulation (ZDES) method when used for secondary flow prediction. This is carried out in light of three issues commonly investigated for hybrid RANS/LES methods (detection and protection of attached boundary layer, emergence, and growth of resolved turbulent fluctuations and accurate prediction of separation front due to progressive adverse pressure gradient). The studied configuration is the first rotor of a high pressure compressor. Three different turbulence modelings (Spalart and Allmaras model (SA), Menter model with (SST) and without (BSL) shear stress correction) are assessed as ZDES underlying turbulence model and also as turbulence model of unsteady RANS simulations. Whatever the underlying turbulence model, the ZDES behaves well with respect to the first two issues as the boundary layers appear effectively shielded and the RANS-to-LES switch is close downstream of trailing edges and separation fronts leading to a quick LES treatment of wakes and shear layers. Both tip leakage and corner flows are strongly influenced by the Navier–Stokes resolution approach (unsteady RANS vs. ZDES) but the underlying turbulence modelling (SA vs. SST vs. BSL) impacts mainly the junction flow near the hub for both approaches. ZDES underlying turbulence model choice appear essential since it leads to quite different corner flow separation topologies and so to inversion of the downstream stagnation pressure radial gradient. Full article