Large-scale Satellite Research Articles

The Detection of Maize Seedling Quality from UAV Images Based on Deep Learning and Voronoi Diagram Algorithms

Assessing the quality of maize seedlings is crucial for field management and germplasm evaluation. Traditional methods for evaluating seedling quality mainly rely on manual field surveys, which are not only inefficient but also highly subjective, while large-scale satellite detection often lacks sufficient accuracy. To address these issues, this study proposes an innovative approach that combines the YOLO v8 object detection algorithm with Voronoi spatial analysis to rapidly evaluate maize seedling quality based on high-resolution drone imagery. The YOLO v8 model provides the maize coordinates, which are then used for Voronoi segmentation of the field after applying the Convex Hull difference method. From the generated Voronoi diagram, three key indicators are extracted: Voronoi Polygon Uniformity Index (VPUI), missing seedling rate, and repeated seedling rate to comprehensively evaluate maize seedling quality. The results show that this method effectively extracts the VPUI, missing seedling rate, and repeated seedling rate of maize in the target area. Compared to the traditional plant spacing variation coefficient, VPUI performs better in representing seedling uniformity. Additionally, the R2 for the estimated missing seedling rate and replanting rate based on the Voronoi method were 0.773 and 0.940, respectively. Compared to using the plant spacing method, the R2 increased by 0.09 and 0.544, respectively. The maize seedling quality evaluation method proposed in this study provides technical support for precision maize planting management and is of great significance for improving agricultural production efficiency and reducing labor costs.

Collaborative last-hop scheduling strategy for large-scale LEO constellation routing

The transmission of data packets from satellites to Earth Stations (ESs) is the last hop of satellite network routing. As the final stage of packet transmission, the use of different scheduling strategies directly affects the throughput of the satellite network. Traditionally, researchers have attempted to enhance network performance by investigating inter-satellite routing protocols or inter-satellite data offloading strategies. However, these approaches have failed to address scheduling issues in the last hop of large-scale Low Earth Orbit (LEO) constellations. In this paper, we present the first Cooperative Last-Hop Scheduling (CLHS) strategy for routing in large-scale LEO constellations based on a bidirectional communication domain. In this strategy, we first utilize the bidirectional communication domain to determine the communication ranges of satellites and ESs. Subsequently, an information flow is established to interact with ESs and satellites. The ESs receive and reconstruct the Information Matrix (IM) from the information flow. Moreover, we propose the Maximum Decision Value Priority (MDVP) algorithm, which takes the reconstructed IM as input and computes the scheduling commands for satellites within the communication range of the ESs. To address the issue of multiple ESs simultaneously scheduling the same satellite, we introduce the Collision Avoidance Algorithm (CAA). Finally, to enhance the data packet transmission efficiency of the scheduled satellites, we propose the Weighted First-In-First-Out (WFIFO) algorithm, which is specifically designed for satellite packet dequeuing. We validate the CLHS through simulations on two satellite constellations: the first-generation Starlink constellation with 4409 satellites and the GW-2 constellation with 6912 satellites. The results show that CLHS can achieve better network throughput than traditional strategies. CLHS provides a novel method of scheduling the last hop in large-scale satellite constellations.

Network Slicing Strategy for Real-Time Applications in Large-Scale Satellite Networks With Heterogeneous Transceivers

Network Slicing Strategy for Real-Time Applications in Large-Scale Satellite Networks With Heterogeneous Transceivers

Bacterial-fungicidal vine disease detection with proximal aerial images

Vine disease detection is considered one of the most crucial components in precision viticulture. It serves as an input for several further modules, including mapping, automatic treatment, and spraying devices. In the last few years, several approaches have been proposed for detecting vine disease based on indoor laboratory conditions or large-scale satellite images integrated with machine learning tools. However, these methods have several limitations, including laboratory-specific conditions or limited visibility into plant-related diseases. To overcome these limitations, this work proposes a low-altitude drone flight approach through which a comprehensive dataset about various vine diseases from a large-scale European dataset is generated. The dataset contains typical diseases such as downy mildew or black rot affecting the large variety of grapes including Muscat of Hamburg, Alphonse Lavallée, Grasă de Cotnari, Rkatsiteli, Napoca, Pinot blanc, Pinot gris, Chambourcin, Fetească regală, Sauvignon blanc, Muscat Ottonel, Merlot, and Seyve-Villard 18402. The dataset contains 10,000 images and more than 100,000 annotated leaves, verified by viticulture specialists. Grape bunches are also annotated for yield estimation. Further, tests were made against state-of-the-art detection methods on this dataset, focusing also on viable solutions on embedded devices, including Android-based phones or Nvidia Jetson boards with GPU. The datasets, as well as the customized embedded models, are available on the project webpage.2

A Flexible Topology Control Strategy for Mega-Constellations via Inter-Satellite Links Based on Dynamic Link Optimization

In large-scale satellite constellations, the efficiency of inter-satellite communication is paramount. Traditional topology control strategies, such as the Manhattan configuration, provide stable links but can result in indirect communication paths, affecting the efficiency of information transfer. This paper addresses this issue by proposing an innovative “3 + 1” dynamic topology control scheme. The scheme retains three static links determined by the relative angular velocity and acceleration while introducing a dynamic link based on distance and angular velocity constraints to optimize the link duration and overall network communication efficiency. To address the complexity of matching dynamic links, this paper introduces an elite strategy-based maximum weighted matching algorithm for general graphs. Compared to traditional greedy algorithms, our proposed algorithm significantly improves the link duration and topological stability. Through simulation experiments comparing communication delays between Xiamen and Los Angeles, our results show that the proposed dynamic link scheme substantially reduces the average delay, enhancing the efficiency and flexibility of inter-satellite communication. This research not only extends the duration of inter-satellite links but also provides new perspectives and methodologies for further studies on inter-satellite topology control strategies.

A heat grid-driven method for generation of satellite observation tasks

Managing numerous observation requirements on a large-scale satellite constellation is arduous for satellite control departments. Concurrently, achieving fast and effective task scheduling necessitates autonomous observation task generation. To address this issue, we proposed a novel method combined with Observation Requirements Heat Grid Model (ORHGM) and a Grid Aggregation Algorithm Based on Field of View Exploration (GAAFVE). Firstly, we define a uniform description of the observation requirements for various target types by devising a heat grid model predicated on geographic grid segmentation. Experimental outcomes reveal that our uniform model saves up to 31.93% of the observation tasks for fulfilling identical requirements by independent task generation for various target types. Secondly, based on this model, an autonomous task generation is completed by GAAFVE. Under the constraints regarding satellite imaging width, our algorithm facilitates heat grid search for non-connected regions and demonstrates superiority in computing time compared to other clustering-based task algorithms. The proposed method can allow managing large-scale observation requirements to be completed relatively quickly for satellite control departments. This method can also be extended to task generation in other multi-agent systems beyond satellites, such as the automatic generation of multi-UAV disaster reconnaissance tasks.

Seasonality of spectral radiative fluxes and optical properties of Arctic sea ice during the spring–summer transition

The reflection, absorption, and transmittance of shortwave solar radiation by sea ice play crucial roles in physical and biological processes in the ice-covered Arctic Ocean and atmosphere. These sea-ice optical properties, particularly during the melt season, significantly impact energy fluxes within and the total energy budget of the coupled atmosphere-ice-ocean system. We analyzed data from autonomous drifting stations to investigate the seasonal evolution of the spectral albedo, transmittance, and absorptivity for different sea-ice, snow, and surface conditions measured during the MOSAiC expedition in 2019–2020. The spatial variability of these properties was small during spring and increased strongly after melt onset on May 26, 2020, when liquid water content on the surface increased, largely accounting for the enhanced variability. The temporal evolution of surface albedo and sea-ice transmittance was mostly event-driven, thus containing episodic elements. Melt ponds reduced the local surface albedo by 31%–45%. Over the melting season, single ponding events increased the energy deposition of the sea ice by 35% compared to adjacent bare ice. Thus, single melt ponds may impact the summer energy budget as much as seasonal evolution over 1 month. Absorptivity and transmittance showed strong temporal and spatial variabilities independently of surface conditions, possibly due to the different internal sea-ice properties and under-ice biological processes. The differences in seasonal evolution shown for different sea-ice conditions strongly impacted the partitioning of shortwave solar radiation. This study shows that the formation and development of melt ponds, in reducing albedo by a third of bare ice sites, can notably increase the total summer heat deposition. The vastly different seasonal evolutions, different sea-ice conditions, and timing and duration of ponding events need to be considered when comparing local in-situ observations with large-scale satellite remote sensing datasets, which we suggest can help to improve numerical models.

BBD: a new hybrid method for geospatial building boundary detection from huge size satellite imagery

AbstractBuildings that are constructed without the necessary permits and building inspections affect many areas, including safety, health, the environment, social order, and the economy. For this reason, it is essential to determine the number of buildings and their boundaries. Determining the boundaries of a building based solely on its location in the world is a challenging task. In the context of this research, a new approach, BBD, is proposed to detect architectural objects from large-scale satellite imagery, which is an application of remote sensing, together with the geolocations of buildings and their boundaries on the Earth. In the proposed BBD method, open-source GeoServer and TileCache software process huge volumes of satellite imagery that cannot be analyzed with classical data processing techniques using deep learning models. In the proposed BBD method, YOLOv5, DETR, and YOLO-NAS models were used for building detection. SAM was used for the segmentation process in the BBD technique. In addition, the performance of the RefineNet model was investigated, as it performs direct building segmentation, unlike the aforementioned methods. The YOLOV5, DETR and YOLO-NAS models in BBD for building detection obtained an f1 score of 0.744, 0.615, and 0.869 respectively on the images generated by the classic TileCache. However, the RefineNet model, which uses the data generated by the classic TileCache, achieved an f1 score of 0.826 in the building segmentation process. Since the images produced by the classic TileCache are divided into too many parts, the buildings cannot be found as a whole in the images. To overcome these problems, a fine-tuning based optimization was performed. Thanks to the proposed fine-tuning, the modified YOLOv5, DETR, YOLO-NAS, and RefineNet models achieved F1 scores of 0.883, 0.772, 0.975 and 0.932, respectively. In the proposed BBD approach, the modified YOLO-NAS approach was the approach that detected the highest number of objects with an F1 score of 0.975. The YOLO-NAS-SAM approach detected the boundaries of the buildings with high performance by obtaining an IoU value of 0.912.

Scheduling observation tasks for large-scale satellite constellation

With the increasing demand for observations and the development of satellite technology, the scale of the Earth Observation Satellites (EOS) constellation is growing. To make the constellation more efficient during operation, scheduling imaging tasks of large-scale constellations is imperative. To solve the scheduling problem, we designed an allocation strategy for the daily operation and a Deep reinforcement learning (DRL) method for in-orbit emergencies. The results show that both of the method is effective.

Deterministic Global 3D Fractal Cloud Model for Synthetic Scene Generation

This paper describes the creation of a fast, deterministic, 3D fractal cloud renderer for the AFIT Sensor and Scene Emulation Tool (ASSET). The renderer generates 3D clouds by ray marching through a volume and sampling the level-set of a fractal function. The fractal function is distorted by a displacement map, which is generated using horizontal wind data from a Global Forecast System (GFS) weather file. The vertical windspeed and relative humidity are used to mask the creation of clouds to match realistic large-scale weather patterns over the Earth. Small-scale detail is provided by the fractal functions which are tuned to match natural cloud shapes. This model is intended to run quickly, and it can run in about 700 ms per cloud type. This model generates clouds that appear to match large-scale satellite imagery, and it reproduces natural small-scale shapes. This should enable future versions of ASSET to generate scenarios where the same scene is consistently viewed from both GEO and LEO satellites from multiple perspectives.

Service blockage on the downlink in large-scale satellite optical networks: a multi-downlink scheduling for routing selection

Over years of space laser communication technology advances, satellite optical networks (SONs) have emerged as a pivotal component in 6 G networks. Satellite services are transmitted from the global view, undergoing transmission through SONs, and being downloaded to the targeted areas. However, the transmission capacity of satellites passing through the areas where users are concentrated may be insufficient to download services transmitted worldwide. This problem exists in various kinds of satellite networks and may cause a large amount of service congestion. In this paper, we propose a multi-downlink delivery routing selection (MDD-RS) strategy to study the total utilization of transmission capacity of SONs. We construct an integer linear programming (ILP) model to establish an optimal case study for minimal network capacity occupation. Also, we design an online option, MDD-RS heuristic algorithm, dynamically calculating path routes, considering bandwidth allocation and resource constraints. A comparative analysis against the conventional single-downlink scheme reveals superior performance of the MDD-RS heuristic algorithm, with a reduction in blocking probability of 0.129 and an improvement in bandwidth utilization of 0.032.

Long-span fiber composite truss made by coreless filament winding for large-scale satellite structural systems demonstrated on a planetary sunshade concept

Climate change necessitates exploring innovative geoengineering solutions to mitigate its effects—one such solution is deploying planetary sunshade satellites at Sun–Earth Lagrange point 1 to regulate solar radiation on Earth directly. However, such long-span space structures present unique technical challenges, particularly structural scalability, on-orbit manufacturing, and in-situ resource utilization. This paper proposes a structural concept for the sunshade’s foil support system and derives from that a component-level modular system for long-span fiber composite lightweight trusses using coreless filament winding. Within a laboratory-scale case study, the component scalability, as well as the manufacturing and material impacts, were experimentally investigated by bending deflection testing. Based on these experimental results, FE models of the proposed structural concept were calibrated to estimate the maximum displacement and mass of the foil support structure, while comparing the influences of foil edge length, orbital load case, and material selection.

Transformer-based semantic segmentation for large-scale building footprint extraction from very-high resolution satellite images

Extracting building footprints from extensive very-high spatial resolution (VHSR) remote sensing data is crucial for diverse applications, including surveying, urban studies, population estimation, identification of informal settlements, and disaster management. Although convolutional neural networks (CNNs) are commonly utilized for this purpose, their effectiveness is constrained by limitations in capturing long-range relationships and contextual details due to the localized nature of convolution operations. This study introduces the masked-attention mask transformer (Mask2Former), based on the Swin Transformer, for building footprint extraction from large-scale satellite imagery. To enhance the capture of large-scale semantic information and extract multiscale features, a hierarchical vision transformer with shifted windows (Swin Transformer) serves as the backbone network. An extensive analysis compares the efficiency and generalizability of Mask2Former with four CNN models (PSPNet, DeepLabV3+, UpperNet-ConvNext, and SegNeXt) and two transformer-based models (UpperNet-Swin and SegFormer) featuring different complexities. Results reveal superior performance of transformer-based models over CNN-based counterparts, showcasing exceptional generalization across diverse testing areas with varying building structures, heights, and sizes. Specifically, Mask2Former with the Swin transformer backbone achieves a mean intersection over union between 88% and 93%, along with a mean F-score (mF-score) ranging from 91% to 96.35% across various urban landscapes.

Latency guaranteed joint optimal traffic intelligent scheduling in large-scale satellite networks

With the continuous growth of the scale and traffic bearing capacity of satellite networks, it becomes increasingly difficult to ensure the traffic end-to-end latency performance in scheduling processes. Inefficient traffic scheduling methods cannot adapt to the dynamic variations of topology connections and traffic loads of the entire network, which results in the timeout of massive transmissions and the waste of limited satellite network resources. In response to the above issues, we analyze the factors affecting the latency performance at the satellite node level and network level from the perspective of traffic packets. On this basis, we construct a Markov decision process (MDP) based on states including residual latency, residual hops, satellite load, and priority. To deal with large amounts of states and unpredictable state transition probabilities in the MDP, we propose a latency guaranteed joint optimal traffic intelligent scheduling algorithm based on the asynchronous advantage actor–critic. The simulation results show that our method improves the packet successful delivery ratio by at least 20% compared to contrast methods, while satisfying packet end-to-end latency requirements.

Dynamic Traffic Grooming Based on Virtualization-Plane-Aided Optimization for Elastic Optical Satellite Networks

With the increase in global wireless traffic, the use of large-scale satellite networking to provide ubiquitous access is one of the essential trends of future 6G network development. Elastic optical satellite networks (EOSNs) are widely considered a flexible solution for future satellite communication. However, with the continuous proliferation of network devices and users, the growing disparity between user demands and the limited bandwidth and capacity of the network is becoming increasingly noticeable. This has led to issues such as constrained network resource utilization and resource fragmentation. Therefore, EOSNs must efficiently address the challenge of allocating scarce bandwidth resources. Effective traffic grooming methods will be applied to EOSNs to solve the problem of bandwidth shortage. This paper proposed a dynamic traffic grooming algorithm based on virtualization-plane-aided optimization (DTG-VPO) to facilitate the bandwidth allocation for EOSNs. Firstly, the nodes of the alternative paths were graded, and the weights of the subsequent hop links were modified. Then, the path was evaluated using link weights, alternative paths were selected in the virtual and physical topologies, respectively, and a path set was constructed. Finally, a resource block evaluation parameter was designed to quantify the quality of candidate resource blocks and rank them. A series of simulations have evaluated the traffic-blocking probability and wavelength utilization under different traffic loads. The link resource was more fully utilized compared with other traffic grooming algorithms. The blocking probability can be reduced by 75%, while wavelength utilization can be improved by 8.1%.

Multi-satellite cooperative scheduling method for large-scale tasks based on hybrid graph neural network and metaheuristic algorithm

Traditional heuristic optimization algorithms are no longer applicable to the multiple satellites scheduling with large-scale tasks, as they are unable to provide satisfactory performance in terms of convergence and speed when addressing complex constraint conflicts. Therefore, we propose a multi-satellite cooperative scheduling method for large-scale tasks based on a hybrid graph neural network (GNN) and metaheuristic algorithm. By using the representation and extraction capability of GNN for relations in graph, the features of large-scale tasks and their constraint relations were expressed, and the generalized knowledge was extracted. And combined with metaheuristic algorithm, the optimization framework for large-scale satellite task scheduling based knowledge and data was implemented. The method consists of a GNN pre-solver module for static constraint conflicts and a metaheuristic optimization module for dynamic constraint conflicts of satellite missions. In the GNN pre-solver module, the graph sample and aggregate network was used to learn and extract the common knowledge in the static task-conflict graph, and provided effective prior information for the metaheuristic optimization module. The quadratic unconstrained binary optimization loss function was designed with multiple influencing factors, to convert the discrete optimization function into a continuous function; a greedy threshold was also added to improve the task completion rate. Finally, numerical experimental results showed that the proposed method can achieve an efficient solution to the multi-satellite scheduling problem with tens of thousands tasks. Compared with the commonly used multi-satellite scheduling algorithms (the genetic algorithm (GA) and greedy-GA), the proposed method can obtain higher-quality solutions under the same conditions and greatly improve the computational efficiency of large-scale mission planning.

Semantic-Aware Jointed Coding and Routing Design in Large-Scale Satellite Networks: A Deep Learning Approach

Semantic-Aware Jointed Coding and Routing Design in Large-Scale Satellite Networks: A Deep Learning Approach

DFSat: Deep Federated Learning for Identifying Cyber Threats in IoT-based Satellite Networks

The integration of satellite systems with smart computing and networking technologies, such as the Internet of Things (IoT), has intensely augmented sophisticated cyberattacks against satellite environments. Resisting cyber threats to complex and large-scale satellite configurations has been enormously challenging, owing to the deficiency of high-quality samples of attack data collected from distributed satellite networks. This study proposes a novel federated learning-based deep learning framework for intrusion detection, named DFSat, to identify cyberattacks from IoT-integrated satellite networks. We develop a distributed deep learning-enabled attack detection method using a recurrent neural network. We then build a federated learning architecture which, utilizes several IoT-integrated satellite networks to preserve the privacy and security of DFSat's parameters throughout the learning process. Extensive experiments have been conducted using communication rounds on an IoT-based network dataset to validate the efficiency of DFSat. The results revealed that the proposed framework significantly distinguishes complex cyberattacks, outperforming recent state-of-the-art intrusion detection techniques, validating its usefulness as a viable deployment framework in IoT-integrated satellite networks.

Enhanced Time Discretization for Temporal Graph-Based Continuous Service Provisioning in Large-Scale Satellite Networks

Enhanced Time Discretization for Temporal Graph-Based Continuous Service Provisioning in Large-Scale Satellite Networks

Satellite Video Multi-Label Scene Classification With Spatial and Temporal Feature Cooperative Encoding: A Benchmark Dataset and Method.

Satellite video multi-label scene classification predicts semantic labels of multiple ground contents to describe a given satellite observation video, which plays an important role in applications like ocean observation, smart cities, et al. However, the lack of a high-quality and large-scale dataset prevents further improvement of the task. And existing methods on general videos have the difficulty to represent the local details of ground contents when directly applied to the satellite videos. In this paper, our contributions include (1) we develop the first publicly available and large-scale satellite video multi-label scene classification dataset. It consists of 18 classes of static and dynamic ground contents, 3549 videos, and 141960 frames. (2) we propose a baseline method with the novel Spatial and Temporal Feature Cooperative Encoding (STFCE). It exploits the relations between local spatial and temporal features, and models long-term motion information hidden in inter-frame variations. In this way, it can enhance features of local details and obtain the powerful video-scene-level feature representation, which raises the classification performance effectively. Experimental results show that our proposed STFCE outperforms 13 state-of-the-art methods with a global average precision (GAP) of 0.8106 and the careful fusion and joint learning of the spatial, temporal, and motion features are beneficial to achieve a more robust and accurate model. Moreover, benchmarking results show that the proposed dataset is very challenging and we hope it could promote further development of the satellite video multi-label scene classification task.