Original high-definition radar data contains rich environmental information, including distance, Doppler velocity, and azimuth. However, extracting robust features from such sparse and noisy frequency-domain data remains a challenge. To address this issue, this paper proposes an improved multi-task network, the Efficient Feature Aggregation with Balanced Attention Radar Network (EFA-RadNet). This network introduces the VoVNetV2 architecture into the field of raw radar perception and effectively preserves feature diversity across different receptive fields through a One-Shot Aggregation (OSA) module, avoiding signal aliasing. In addition, we propose an attention mechanism module, Balanced effective Squeeze-Excitation (B-eSE), which is better suited for sparse radar processing and effectively addresses the problem of weak target loss in the radar spectrum. Experiments on the RADIal dataset show that our EFA-RadNet achieves excellent target detection performance while also attaining optimal accuracy in free space segmentation.

In Global Navigation Satellite Systems (GNSS), high-precision position coordinates are typically determined by establishing a double-difference carrier phase observation model and resolving the integer ambiguities within it. Therefore, the ability to fix integer ambiguities rapidly and accurately is a critical challenge in carrier phase measurements. To address the problem of double-difference integer ambiguity, this paper proposes a Hybrid Adaptive Differential Evolution Grey Wolf Optimizer (HADE-GWO) algorithm. Comparative experiments focusing on computation speed and stability were conducted against the GWO, LAMBDA, and M-LAMBDA algorithms. The results show that while achieving the same fixing success rate as the LAMBDA and M-LAMBDA algorithms, the HADE-GWO algorithm finds the optimal ambiguity solution in less time. To validate the high-dimensional ambiguity resolution capability of the HADE-GWO algorithm, 6-dimensional and 12-dimensional integer ambiguity resolution tests were performed. The outcomes indicate that the HADE-GWO algorithm possesses excellent high-dimensional resolution capabilities. Finally, an application experiment was conducted using single-frequency data from GPS and BeiDou (BDS) systems. The results demonstrate that the algorithm can achieve centimeter-level positioning accuracy in a combined single-frequency GPS+BDS solution.

Objective Object detection, as one of the core technologies in computer vision, has been widely applied in fields such as autonomous driving, security surveillance, and remote sensing image analysis. However, current mainstream object detection technologies are primarily based on single-modal images such as visible light or infrared, which suffer from insufficient detection accuracy in complex environments like nighttime or low-light conditions, making it difficult to meet practical demands. Given that visible light images can effectively capture detailed texture information of targets under sufficient lighting, while infrared images excel at capturing target contours in low-light or nighttime scenarios, effectively fusing these two modalities has become the key to addressing the aforementioned challenges. Existing research mainly focuses on leveraging the similar features between the two modalities to enhance detector performance, but tends to overlook the complementarity between modalities and the full representation of their inherent characteristics. Therefore, how to effectively enhance the intrinsic feature representation within each modality and fully exploit the complementary features between modalities remains a key issue to be addressed for further improving the performance of object detectors. Methods To fully exploit the inherent characteristics of visible and infrared modalities as well as their complementary features, this paper proposes a cross-modality object detection algorithm based on cross-attention feature enhancement, named CAMNet. First, the baseline model is extended to a dual-stream input structure to separately extract features from visible and infrared images. Next, two plug-and-play core modules are designed: the feature enhancement module (FEM) and the cross-attention module (CAM). Specifically, FEM leverages a self-attention mechanism to enhance the intrinsic feature representation within each modality, effectively suppressing noise interference. To reduce model complexity, FEM also integrates a spatial feature aggregation (SFA) module, which efficiently aggregates features using an adaptive mixed pooling technique, thereby improving network efficiency. The CAM innovatively employs a Re-Softmax function and utilizes a cross-attention mechanism from both channel and spatial dimensions to effectively extract complementary features between modalities. Finally, the FEM and CAM modules are embedded into each feature extraction stage of the backbone network, significantly enhancing the model's ability to fuse and utilize dual-modality feature information. Results and Discussions To fully validate the effectiveness of the proposed method in this paper, comprehensive experiments are conducted on the VEDAI and FLIR datasets. The results show that on the VEDAI dataset, compared to the baseline method, CAMNet improves precision, recall, and mAP50 (mean average precision at intersection over union threshold of 0.50) by 4.2, 6.4, and 6.4 percentage points, respectively (Table 2); compared to other state-of-the-art methods, CAMNet also demonstrates significant advantages (Table 3). On the FLIR dataset, CAMNet exhibits advanced performance, with its mAP50 increasing by 3.3 percentage points compared to the baseline method (Table 7). Qualitative analysis further reveals that in scenarios involving similar objects and small objects, CAMNet achieves higher precision and recall compared to the baseline method (Fig. 6), indicating that compared to the Add fusion strategy, the proposed method can more effectively mine and utilize modality-specific features and cross-modal complementary features. Furthermore, in complex scenarios such as occlusion, dense objects, and small objects, CAMNet shows higher detection precision and robustness compared to other state-of-the-art methods (Figs. 7-8), fully demonstrating the superiority of the proposed method. Ablation experiments confirm that there is no conflict between the FEM and CAM modules, and they can effectively collaborate to jointly enhance the overall performance of the model (Table 4). Conclusions This paper addresses the issues of insufficient intra-modal feature representation and inadequate inter-modal feature fusion in existing bimodal fusion detection algorithms by proposing a novel fusion detection method named CAMNet. The method designs two efficient plug-and-play modules: FEM and CAM. While balancing model performance and complexity, CAMNet demonstrates excellent cross-modal object detection capabilities. Specifically, the FEM focuses on mining and strengthening the representational power of inherent features within a single modality. In contrast, the CAM is dedicated to extracting complementary features between modalities, refining and enhancing them in both the channel and spatial dimensions, respectively. By embedding these modules into the baseline model, the network's joint modeling capability for both inherent features and complementary features is significantly enhanced. Experimental validation on the VEDAI and FLIR datasets demonstrates that CAMNet achieves mAP50 of 85.9 degrees o and 77.4 degrees o, respectively, significantly surpassing both the baseline model and current state-of-the-art methods. Furthermore, the proposed method maintains robust performance even in complex scenarios involving occlusion, truncation, and dense objects, showcasing excellent robustness and detection accuracy.

Millimeter-wave automotive radar is essential for autonomous driving due to its robustness under adverse weather conditions. However, radar point clouds are inherently sparse, noisy, and of low resolution, posing significant challenges for object detection. Graph neural networks (GNNs) are a natural fit for radar point clouds, yet conventional approaches rely on static, heuristic graph constructions such as fixed $k$ -nearest neighbors ( $k$ -NN) that ignore radar-specific multi-attribute cues including Doppler velocity and radar cross-section (RCS). This paper presents Adaptive Graph Construction Learning (AGCL), a framework in which a positive semidefinite (PSD) Mahalanobis metric is learned offline to reweight edges on a sparse candidate graph, yielding feature-aware connectivity. The metric learning is formulated by minimizing a graph Laplacian regularizer and is performed as a one-time preprocessing step for each point-cloud graph; the resulting edge weights are cached and reused during detector training and inference, incurring low online overhead. A block-wise optimization strategy enables scalability to large graphs. Two instantiations are considered: a full PSD metric (AGCL-L) evaluated on RadarScenes and a constrained diagonal variant (AGCL-D) for lightweight plug-in integration on View-of-Delft. The learned feature graph is integrated into downstream detectors either as edge weights for GNN message passing (AGCL-Edge) or as neighborhood-statistics feature augmentation (AGCL-Stats), requiring only interface-level changes rather than backbone redesign. On RadarScenes, AGCL improves over a strong RadarGNN baseline; additional gains are also validated on View-of-Delft using modern detector families including RadarPillars and LEROjD/DSVT. Ablation studies confirm the benefit of incorporating Doppler and RCS cues.

Phase measurements from the global navigation satellite system are fundamental to high-precision applications. However, in complex observation environments, these signals are susceptible to adverse propagation conditions, such as ionospheric scintillation and multipath effects. Adverse propagation conditions induce frequent cycle slips, causing discontinuities in carrier phase measurements. To this end, an adaptive cycle slip detection method is introduced, which is specifically tailored for ionospheric scintillation environments. Our approach integrates a grey wolf optimizer-long short-term memory model with geometry-free (GF) combinations to establish a dynamic detection threshold, which is informed by a predicted scintillation amplitude index. The method was validated using real-world BeiDou navigation satellite system (BDS) and global positioning system (GPS) data, where it demonstrated a significant reduction in false alarm rates compared with traditional GF combination methods. Specifically, false alarm rates decreased by over 30% (up to 83.3%) for BDS data and over 20% (up to 60.0%) for GPS data across various sampling rates.

Automotive millimeter-wave radar produces sparse point clouds with Doppler velocity and radar cross-section (RCS), but graph detectors typically use a shared representation for semantic prediction and box regression despite their different propagation requirements. We propose multi-GSO spectral filtering (MGSF), a residual module that filters radar features over geometry-, Doppler-, and RCS-defined graph shift operators and fuses diffusion and residual components with a node-adaptive gate. MGSF-TD applies full multi-GSO refinement to semantic prediction and geometry-only refinement to box regression. On the complete RadarScenes validation set, MGSF-TD improves the official RadarGNN checkpoint from 60.19 to 60.59 mAP and from 74.06 to 75.10 mean foreground F1 (FG-F1). Across three MGSF-TD training seeds, the FG-F1 margin under RCS noise increases from +1.15 at 3 dBsm to +2.38 at 20 dBsm; seed-42 full-validation mAP margins are +0.24, +1.07, and +1.82. Controls show that geometry-only diffusion explains part of the gain and the RCS operator contributes most clearly at low-to-moderate noise, whereas a parameter-matched widened baseline matches or exceeds MGSF-TD under severe RCS and Doppler corruption. Cross-sensor diagnostics reproduce the Doppler failure trend but not the severity-dependent RCS gain. MGSF-TD therefore offers a balanced, physically interpretable operating point rather than a universal robustness gain.

To address the issues of insufficient noise suppression and false detections in the FFTRadNet model for target detection in DDMA-MIMO millimeter-wave radar and free-space segmentation in autonomous driving, this paper proposes an improved model, CAN-FFTRadNet, which integrates a cascaded attention mechanism with the lightweight MobileNetV3 network. This model replaces the original feature extraction network with MobileNetV3, effectively reducing the number of parameters and computational complexity. It combines the ECA + SA cascaded attention mechanism to enhance both channel and spatial feature representations. Additionally, the Weighted Feature Fusion with Sparse Coding (WFF-SC) method is introduced for multi-scale feature fusion to improve integration efficiency. Experimental results show that in the tasks of object detection and free-space segmentation, the CAN-FFTRadNet method achieves improved recall rate and segmentation quality while maintaining comparable precision. Specifically, its Average Recall (AR) and mean Intersection over Union (mIoU) are increased by 7.1

Light Detection and Ranging (LiDAR) three-dimensional (3D) object detection degrades under point sparsity, outliers, coordinate noise, and calibration drift, yet detector evaluation remains largely limited to clean benchmarks. This study focuses on sensing robustness rather than detector redesign. We introduce Bounded Graph Conditioning (BGC)—a deterministic pre-voxelization front-end that applies k-nearest-neighbor (kNN) neighborhood averaging with bounded residual correction upstream of an unchanged detector backbone. BGC is evaluated together with a reproducible sensor-degradation stress protocol and a risk-constrained operating-boundary analysis. Experiments on KITTI with PointPillars, SECOND, and Voxel R-CNN show that BGC most clearly improves retained detection quality and feasible operating coverage under strong noise and strong outlier stress; gains under other degradation types are smaller and backbone-dependent. In the primary score-level box-disjoint calibration/test evaluation on SECOND, maximum feasible coverage at a target risk bound of 0.2 improves from 0.0754 to 0.1374 under strong noise (σ=0.10 m) and from 0.1323 to 0.1591 under strong outliers (p=0.10); a cross-backbone check on Voxel R-CNN confirms the same direction (0.1860→0.2864). Comparison with traditional filtering (SOR and ROR) reveals complementary strengths across fault types. A range-adaptive BGC variant that adjusts parameters per distance bin further improves performance under mixed unknown faults, spherical-coordinate noise, and on a dataset-matched nuScenes validation (adaptive BGC mAP/NDS: 0.2687/0.4493 vs. baseline 0.2471/0.3846 under strong noise). Severe translation drift collapses all configurations to full rejection, exposing an explicit sensing boundary beyond the reach of local conditioning. These results support BGC as a practical sensor-side robustness enhancement under the studied degradation protocol, with conditional rather than universal applicability across backbones and fault types.

Roadside perception systems, also known as roadside units (RSUs), are critical in Vehicle-to-Everything (V2X) applications, yet spatio-temporal asynchrony between multiple sensors severely compromises the accuracy of fusion. In this paper, a spatio-temporal synchronization method for millimeter-wave (MMW) radar and camera fusion is proposed, integrating target matching based on dynamic time warping (DTW) with spatio-temporal parameter estimation. Leveraging the advantages of DTW in time-series alignment to calculate the similarity between radar and visual trajectories enables target matching and parameter estimation in sparse scenes. This method was validated on a real-world dataset containing over 30 pedestrian trajectories, covering scenarios with varying densities ranging from one to six pedestrians. The results indicate a temporal offset of 0.116 s between the camera and radar. Following synchronization, the average spatial deviation decreased from 1.4358 to 0.1074 m in the x-direction (i.e., across the road) and from 3.0732 to 0.1775 m in the y-direction (i.e., along the road). Consequently, this method provides an efficient solution for deploying roadside perception systems in sparse traffic environments.

GNSS array receivers suffer tracking degradation under array nonidealities such as element-position perturbations, channel amplitude/phase errors, and slowly varying manifold mismatch. Conventional blind anti-jamming suppresses interference, but adaptive weight fluctuations can propagate into the correlator domain, increasing cross-branch correlation, causing Early/Late metric imbalance, and reducing Prompt phase consistency. Existing noncoherent combining methods mainly convert multi-branch correlator outputs into scalar energy metrics for code tracking, leaving the carrier loop's complex Prompt input insufficiently constrained. To address this problem, we propose a blind adaptive joint code-carrier channel-combining method for nonideal arrays. After first-stage anti-jamming, the method estimates an Early/Late correlator-domain covariance matrix and reuses it as a shared statistical constraint. In the code loop, this matrix drives whitened noncoherent energy combining with closed-loop gain normalization to stabilize the DLL discriminator scale. In the carrier loop, it is combined with a Prompt-derived coherent direction to form a covariance-constrained PLL complex input. Simulations under wideband interference, static array errors, and dynamic mismatch show that the proposed J-WNCC reduces both code-phase error and carrier-phase jitter, improving joint tracking robustness in nonideal array environments. Ablation results further reveal a dominant-effect separation: DLL gain normalization mainly calibrates the whitened code-discriminator scale, whereas coherent Prompt combining mainly reconstructs the complex PLL input.

Spoofing detection is critical for GNSS security. To address the issues of low detection rates and insufficient coverage in traditional methods, this study proposes an eye diagram detection method based on the multiscale Canny algorithm with minimum misjudgment probability (EDDM-MSC-MMP). Unlike conventional correlation peak distortion detection techniques, the proposed method uses the MSC-MMP algorithm to perform multiscale edge extraction from the eye diagram generated from the receiver’s correlation values. It then calculates the image threshold using minimum misjudgment probability to ensure the accuracy of the eye diagram’s edges. This enables the accurate detection of subtle changes in the eye diagram, leading to the better identification of spoofing signals. The results show that the MSC-MMP outperforms traditional edge extraction algorithms by over 0.072 in terms of the optimal dataset scale F score (ODS-F). Compared to signal quality monitoring (SQM) and Carrier-to-Noise Ratio methods, the EDDM-MSC-MMP method increases spoofing detection coverage by over 60%, achieving the highest detection rate in the TEXBAT dataset. Overall, the EDDM-MSC-MMP method improves the reliability and coverage of spoofing detection, providing an effective solution for GNSS spoofing detection.

Spoofing intrusions pose a major threat to user security by delivering incorrect information. The detection rate of existing signal quality monitoring (SQM) metrics notably decreases when faced with numerous specific combinations of code phases and carrier phases in spoofing signal instances. To increase the detection rate and coverage, we exploit the offset detection capability of different correlators and propose metrics: multipoint slope differential (MuSD) and multipoint slope differential averaging (MuSDA). In addition, this paper proposes a Weighted Moving Average Bias Correction (WMA-BC) algorithm for metric post-processing. Comparative experiments with Moving Average (MA) and Moving Variance (MV) based SQM methods demonstrate that the WMA-BC algorithm achieves substantial advantages in Detection Rate enhancement and significantly improves the Area Under Curve (AUC) of the Receiver Operating Characteristic (ROC). In experiments with different code and carrier phase offsets, the detection coverage of the MuSD and MuSDA metrics reached 96.1% and 95.8%, respectively, which are much greater than those of other metrics. From the detection rates obtained in seven spoofing intrusion experiments based on the Texas Spoofing Test Battery (TEXBAT) dataset collected at the University of Texas, the proposed MuSDA and MuSD metrics outperform other metrics by approximately 11% to 97%.

Currently, single-sensor data is frequently utilized in technologies such as object detection. However, in certain scenarios, some sensors may experience failure or information loss, significantly impacting model performance. Given the notable complementarity between infrared and visible images at the information level, it is helpful to improve the robustness and reliability of the model by fusing them and applying them to object detection and other technologies. Nevertheless, most prevalent infrared and visible image fusion methods focus on exploring invariant features across multimodal images, somewhat neglecting the inherent characteristics of the images themselves, leading to issues like structural blurriness and unclear detailed textures in the fused images, which fail to meet application demands. To overcome this challenge, this paper proposes an Infrared and visible image fusion network based on multistage progressive injection, termed MPIFusion. To effectively leverage the inherent characteristics and complementarity of images, and address the issues of structural blurriness and unclear detailed textures in fused images, we introduce a Dual-channel Shallow Detail Fusion Module (DC-SDFM) and a Deep Feature Fusion Block (DFFB). These modules first enhance the original features and then fuse the hierarchical features of infrared and visible images with the aid of an attention mechanism module. Furthermore, we construct a progressive injection layer based on the Information Fusion Module (IFM), integrating the fused features within the same framework to generate high-quality infrared and visible fused images. Extensive experiments demonstrate that our MPIFusion outperforms 15 existing fusion methods in terms of performance. The generated fused images not only highlight global and local detail features but also exhibit higher clarity and contrast. Finally, we apply fusion methods to object detection scenarios, and the results show that MPIFusion exhibits significant superiority in such scenarios, providing more robust and reliable image support. The source code is available at https://github.com/Kaixuan-Chang/MPIFusion .

Due to the narrow nulls formed by the Power Inversion (PI) algorithm, it fails to suppress jamming signals in highly dynamic scenarios effectively. This paper proposes a null broadening algorithm based on eigenvalue sorting. Unlike other algorithms, this one does not require prior knowledge of the direction of jamming. It is based on the Covariance Matrix Taper (CMT) algorithm, which orders the eigenvalues of the sampling covariance matrix. The sample covariance matrix’s eigenvalues are sorted to provide new sample data, and the rebuilt covariance matrix is then averaged forward and backward. The experimental results demonstrate that the proposed algorithm can effectively broaden the null. Compared with the CMT algorithm, the null in the jamming direction is, on average, approximately 22 dB deeper under the experimental conditions, and the gain in the direction of the sound signal is increased by around 15 dB. Moreover, the signal can be successfully acquired even when the input jamming-to-signal ratio (ISR) is relatively low. When there is a deviation in the jamming direction, the proposed algorithm demonstrates robust null broadening performance even with a small number of snapshots. The output SINR of the proposed algorithm exhibits a nearly linear relationship with the input SNR.

To investigate the influence of scallop farming on the biogeochemical characteristics of colloidal organic matter (COM, 1 kDa-0.7 μm), surface and bottom seawater samples collected from a bay scallop mariculture area (MA) and its adjacent waters were size-fractionated and analyzed for absorption and fluorescence characteristics. Compared to inshore shallow water area and non-mariculture area (NMA), COM in MA exhibited the highest proportion of a350 in bulk dissolved organic matter, while the contribution of its 100 kDa-0.7 μm high molecular weight fraction (HCOM) was the lowest. Protein-like components, including tryptophan-like C1 and tyrosine-like C2, predominated in the fluorescent substances of HCOM; while humic-like components, including microbial humic-like C3 and terrestrial humic-like C4, dominated in the fluorescent substances of 1-3 kDa low molecular weight fraction of COM (LCOM). COM transformation was influenced by scallop farming via selective filter-feeding and enhanced degradation of microorganisms. Compared with NMA, phytoplankton production mainly affected the a350 and protein-like substances of HCOM in the surface seawater of MA, and its contributions increased by 2.4% (a350), 5.6% (C1), and 1.8% (C2) respectively. Meanwhile, microbial degradation significantly influenced component C1 of HCOM in the bottom seawater of MA, reducing its contribution by 33.2%; its impact on the contributions of humic-like fluorescent substances of LCOM were decreased by 2.7% for C3 and 1.1% for C4. These variations suggest that scallop farming promotes the production and transformation of protein-like substances in HCOM (mainly C1), potentially leading to an accumulation of humic-like substances in LCOM due to altered microbial degradation dynamics. Photodegradation, particulate organic matter settling, sediment release and colloidal aggregation/disaggregation also influence the transformation of size-fractionated COM.

By integrating ultraviolet (UV) photocatalytic oxidation digestion with segmented continuous flow analysis technology, an online measurement method and analysis system for the alkaline chemical oxygen demand (CODMn) in seawater, based on the color-change reaction of potassium permanganate, has been established. This represents the first application of UV photocatalytic oxidation technology in the measurement of CODMn in seawater. The system effectively overcomes the limitations of high-temperature and high-pressure digestion methods employed in traditional CODMn analysis. The detection limit was 0.13 mg/L, the linear range was 0.00 mg/L ~ 4.60 mg/L, with a relative standard deviation of 2.30 % (n = 11), and recovery rates of 95.33 % ~ 109.50 %. Additionally, the system minimized interference from salinity, making it suitable for CODMn determination in seawater. Compared to the national standard method, the maximum average relative error was 0.042, lower than the conventional standard of 0.10, indicating high accuracy. The analysis of seawater samples from the Yangma Island area further confirmed the reliability of the system in practical testing and regional organic pollution assessment. The system demonstrated excellent anti-interference capabilities, rapid detection speed, and ease of operation, providing significant application potential in seawater CODMn measurement.

In order to address the issues of velocity ambiguity and target energy dispersion in DDMA-MIMO radar, an enhanced velocity defuzzification methodology is presented. Firstly, the repeated distribution of the target's spectrum is exploited to compress the range Doppler spectrum in the Doppler dimension. This results in the conversion n of the two-dimensional detection to the detection of a single distance-dimensional vector. Furthermore, the subarea peak search is employed to replace the Doppler-dimensional CFAR detection within the Doppler dimension, thereby significantly enhancing the detection efficiency. In addition, a DDMA waveform free of empty bands is designed to estimate the target angle while achieving velocity deblurring through phase compensation and digital beam forming algorithms. The efficacy and superiority of the proposed method are demonstrated through a verification process using measured data.

The collaborative control algorithms of multiagents system have been applied to many Internet of Things (IoT) devices. The anonymous flocking algorithm of multiagents is a core technology in the collaborative control research of multiagents system. It does not need agents to distinguish other agents and obstacles, but most existing researches have specific constraints on the obstacle shape, which limits its practical applications. The obstacle boundary points contain the shape characteristics of the obstacle. To relax the obstacle shape constraint, we assume that all agents can only perceive the position of obstacle boundary points within their sensing radius, and propose an anonymous flocking algorithm with obstacle avoidance via the position of obstacle boundary point. In this algorithm, the consensus term is divided into velocity consensus term and velocity unit direction consensus term. The velocity consensus term is designed to tow agents that perceive the obstacle boundary points preventing them from following to bypass obstacles, and the velocity unit direction consensus term is designed to achieve the matching of velocity unit direction. Additionally, the gradient-based term is designed to realize the separation and aggregation between agents and obstacle boundary points, and the navigational feedback term is designed to lead all agents to realize the group objective following. Furthermore, it is verified through simulations that the proposed algorithm can relax obstacle shape constraint and has better environmental adaptability.