This paper develops an analytical framework for time-correlated shadowed Rician (SR) channels in low-Earth orbit (LEO) satellite downlinks based on a two-snapshot model. The proposed formulation explicitly captures the distinct temporal correlations of the diffuse and line-of-sight (LOS) components, together with the deterministic LOS phase evolution induced by orbital geometry. Based on this model, a closed-form expression is derived for the temporal power correlation coefficient, and power-series representations are obtained for the joint cumulative distribution function (CDF) and probability density function (PDF) of the two-snapshot received power. These results are then used to characterize the joint outage probability (JOP) over short lags. The framework is further extended to sampled outage dynamics by deriving power-based level crossing rate (LCR) and average fade duration (AFD) directly from the marginal and joint power CDFs, without requiring derivative-based continuous-time analysis. Numerical and Monte Carlo (MC) results verify the accuracy of the developed expressions and show that the shadowing level affects not only the magnitude but also the temporal structure of the received-power correlation. The results also reveal how shadowing severity, lag, sampling period, and target data rate jointly govern outage occurrence, recovery, and persistence in realistic LEO downlinks.
Radar small-target detection in complex low-altitude environments remains challenging because weak target echoes are easily masked by heterogeneous clutter, multipath returns, and environment-dependent interference. To improve target–clutter separability, this paper proposes a background-aware multimodal framework that jointly exploits range–Doppler signal structure information (SSI) and satellite-derived environmental priors. Instead of making decisions from isolated range–Doppler cells or directly fusing radar and map images, the proposed method reformulates detection as a target-centric structural discrimination problem. Candidate locations are first generated by a low-threshold CFAR detector, and local SSI patches are extracted to preserve target-induced range–Doppler morphology. A physics-inspired SSI enhancement module is designed to encode log-compressed magnitude, local background contrast, and range- and Doppler-domain structural gradients. Meanwhile, spatially aligned satellite images and land-cover masks are used to construct environmental priors for characterizing local clutter conditions. An enhanced gated cross-modal fusion module then selectively injects background information into radar SSI features while preserving radar features as the dominant representation. Finally, a multi-scale Transformer-based detection head with cross-scale interaction is employed to jointly model fine-grained SSI morphology and broader contextual consistency. Experiments on measured radar data demonstrate that the proposed method outperforms conventional CFAR, representative CNN-, Transformer-, and graph-based detectors, as well as a previous radar–satellite fusion method, especially in low-false-alarm operating regions. Ablation studies, information-scale experiments, and SCIR-level analysis further verify the effectiveness of the proposed components and the importance of appropriate SSI and environmental context scales.
In recent years, adaptive fusion weight assignment technology in distributed multi-sensor networks has garnered significant attention, mainly due to its ability to significantly improve multi-target tracking accuracy and system robustness. However, in sensor networks, when the network topology is unknown or dynamically changing, the incompleteness of information propagation and irrationality of weight assignment often lead to a decline in fusion accuracy and degradation of system performance. Existing mainstream fusion methods typically optimize weights based on fusion performance without considering the different performance characteristics of different sensors. In network environments where sensor performance differences are significant, this may lead to degraded fusion accuracy. In addition, when targets operate within detection areas composed of multiple sensors, traditional methods struggle to accurately determine the target's disappearance status, which often leads to misjudgment issues. To address these challenges, we introduce a flooding algorithm to achieve topology discovery and information propagation of distributed sensor networks and propose an adaptive fusion weight assignment strategy based on detection probability and clutter rate, which uses scaling factors to dynamically adjust fusion weights and realizes information fusion through the weighted arithmetic averaging (WAA) method. Additionally, we establish a fusion extinction model that combines the target existence probabilities from multiple sensors to determine whether the target has disappeared. Meanwhile, the random matrix (RM) model is adopted to describe the target's motion state and shape features, thereby optimizing state estimation. This proposed method can respond promptly to changes in the sensor's working environment and optimize the estimation process. Simulation results show that the proposed algorithm exhibits significant improvements in the accuracy of target state and shape estimation compared with the traditional uniform weight method. It provides an efficient solution especially in cases of network topology changes and fluctuations in detection probability.
Surface exploration of asteroids is a critical yet challenging aspect of deep-space missions. This paper presents the design and capabilities of the Autonomous Retractable-Spoke-Wheel Robot (ARSWR), a mobile robotic system specifically developed for asteroid surface exploration. The retractable spoke-wheel mechanism generates significant normal force through dynamic ground interaction, enabling reliable surface contact under microgravity conditions. The spokes can autonomously adjust their length, allowing the wheels to change diameter and adapt to uneven, rugged terrain for stable locomotion. Additionally, simulation results confirm the feasibility of ARSWR performing hopping maneuvers by either rapidly rotating its wheels or extending its spokes in the asteroid surface—key strategies for traversing complex asteroid landscapes. During mid-air hops, the ARSWR is capable of adjusting its attitude to ensure controlled landings. The system also features autonomous navigation and path planning, utilizing the Point-LIO algorithm for localization. Equipped with a LiDAR sensor and an RGB-D camera, the robot constructs local 3D terrain maps and detects obstacles in real-time, thereby enhancing mission autonomy and improving the overall efficiency of asteroid surface exploration.
Interrupted sampling repeater jamming (ISRJ) is a coherent jamming technique combining both deception and suppression capabilities, posing a serious threat to pulse-Doppler radar. Although existing studies have proposed various waveform design methods to counter the ISRJ, these approaches often struggle to balance jamming suppression performance and Doppler tolerance, limiting their practical application in high-speed target detection scenarios. To address this issue, a joint waveform and filter design method is proposed based on a time-domain masking principle. Precisely, the operational waveform and the operational filter adopt a linear frequency modulation (LFM) scheme to preserve Doppler tolerance, meeting the detection requirements for moving targets; whereas the masking waveform and the masking filter are designed using phase-coded sequences. An objective function is formulated by combining the jamming integration energy and the squared absolute deviation of the pulse compression peak gain from predefined thresholds. Further, a joint optimization framework based on Block Coordinate Descent (BCD) and Majorization-Minimization (MM) are employed to iteratively solve for the masking waveform and filter. In addition, the Lanczos method is introduced to reduce the computational burden of the proposed algorithm. Finally, simulation results demonstrate that the proposed joint design algorithm effectively suppresses ISRJ while maintaining satisfied Doppler tolerance. Compared with existing methods, the designed waveform and filter exhibit superior performance in both anti-jamming capability and Doppler adaptability.
This paper addresses the cooperative guidance problem of intercepting a high-speed high-maneuverability target with multiple UAVs. A zero-effort-miss (ZEM) based relay pursuit method is proposed, which coordinates the UAVs by dynamically selecting an active pursuer and designing its pursuit strategy. Specifically, a straight pursuit strategy is designed for the active pursuer to herd the target, thereby reducing the ZEM of the subsequent pursuer. The optimal switching time between pursuers is derived analytically by constructing two circular boundaries. Although the theoretical analysis utilizes several simplifying assumptions to obtain closed-form solutions, extensive numerical simulations demonstrate that the method is robust and remains effective under more general conditions. The results consistently show that the proposed cooperative scheme achieves a significantly smaller miss distance—and often a successful intercept—in scenarios where a single pursuer fails. This validates the critical advantage of multi-UAV coordination in overcoming the limitations of a single interceptor.
Remote single-terminal sites, such as research stations, vessels, and field deployments, increasingly move operational data over commercial low Earth orbit (LEO) links under time constraints, where the practical question is how a stated service envelope maps a payload and deadline into an isolated-transfer completion-time proxy. We present a measurement-grounded service-envelope analysis of deadline feasibility, defined as the probability that this isolated transfer completion-time proxy meets the stated deadline, over a continuous payload–deadline plane for radio-frequency (RF)-only, optical-only, and hybrid RF/optical access. From 441 hours of retained observations on a residential Starlink terminal, we construct two RF service envelopes: a conservative background-service floor and a logged active-test reference whose median rate is about 234× the engaged-floor median; a reconstructed 10 s grid shows zero instrumentation gaps, and a conservative primary-domain excluded-bin replay changes deadline success by at most 0.35 percentage points. Under explicit RF service envelopes, the measured floor gives weak RF-only proxy support for the selected anchors and the supporting i.i.d.-service FCFS baseline yields median sojourn times above the corresponding deadlines even below the per-workload overload threshold, while the active-test reference is a short-test high-load envelope in which 99.97–99.99% of positive-download tests exceed the anchor required rates under the isolated-job mapping; conservatively counting unsuccessful and zero-download attempts as non-exceedances gives all-attempt lower bounds of 92.84–92.86%. Optical access is effective-availability gated, and measured-floor hybrid fallback gives every simulated job a finite completion time under the adopted abstraction. The resulting feasibility maps characterize envelope-dependent payload–deadline regions; controlled sustained-transfer and transport-level validation remain future work.
To address the limitations of existing low-rank tensor methods in modeling high-order background evolution and their reliance on time-consuming iterative Singular Value Decomposition (SVD), this article proposes a Hypergraph Manifold Regularized Tensor (HMRT) model. Specifically, we first construct a spatio-temporal hypergraph to explicitly model the high-order manifold structure of image sequences, utilizing KD-tree acceleration to efficiently capture complex non-linear correlations. Subsequently, a Hypergraph Laplacian Regularization term is introduced to impose a rigorous smoothness prior on the background component, facilitating the spectral decoupling of the smooth background manifold from high-frequency sparse target singularities. Furthermore, to expedite optimization, we derive a closed-form spectral projection operator. This operator transforms background recovery into a fast spectral filtering process, bypassing expensive iterative SVD steps and reducing computational complexity to near-linear time. Comprehensive experiments on diverse datasets demonstrate that the proposed method significantly outperforms state-of-the-art baselines in both detection accuracy and computational efficiency.
Reconstructing the three-dimensional (3D) structure of noncooperative space targets from inverse synthetic aperture radar (ISAR) image sequences is critically challenging in the field of space situational awareness (SSA). The complex electromagnetic scattering involved in ISAR imaging makes reliable 3D reconstruction difficult and complicates ISAR image rendering from limited-view sequences. To address these challenges, this study proposes ISAR Gaussian Splatting (ISAR-GS), a method for the 3D reconstruction and viewpoint rendering of space targets from ISAR image sequences. A set of Gaussian primitives is used to represent the geometry and scattering properties of space targets. To account for ISAR-specific imaging physics, a differentiable ISAR Gaussian renderer is introduced to jointly model energy attenuation and scattering intensity accumulation. A gradient propagation scheme is further proposed to propagate supervision from rendered images to the Gaussian representation, enabling the joint optimization of geometric parameters and scattering attributes. On the three evaluated simulated datasets, ISAR-GS achieves better 3D reconstruction and viewpoint-rendering results than ISAR-NeRF, a representative implicit neural rendering method, while requiring substantially less training time and enabling faster inference. It also exhibits greater robustness than ISAR-NeRF under the evaluated projection-vector perturbations and echo SNR levels. Furthermore, the practical applicability of the proposed method is verified on a measured ISAR image sequence of the Tiangong-1 space station.
This paper presents a coordinated guidance and control scheme for the fixed-time path tracking of a hypersonic glide vehicle subject to model uncertainties and prescribed performance constraints. The scheme features a prescribed performance guidance law that confines tracking errors within predefined time-varying bounds, explicitly shaping the transient and steady state behavior. Building on this transformed system, a fixed-time nonlinear controller is designed to guarantee convergence within a uniform time independent of initial conditions. To reconcile accurate online adaptation with real time feasibility, a radial basis function neural network is simplified via a minimum learning parameter technique and further enhanced by a composite learning strategy. This strategy incorporates the prediction error from a parallel model together with the tracking error, accelerating convergence while maintaining robustness. An auxiliary compensation mechanism is introduced to mitigate the errors induced by the command filter. The stability of the closed-loop system is established through Lyapunov analysis. Simulation results demonstrate that the proposed approach achieves faster transient convergence, smaller overshoot, and a more favorable trade off between tracking accuracy and computational efficiency compared with conventional neural network based schemes and approaches without prescribed performance.
This study explores a distributed collaborative control framework for an on-orbit assembly mission involving a mixed fleet of rigid and flexible spacecraft. To circumvent the inherent issues of singularity and non-uniqueness in some attitude parameterizations, the spacecraft orientation is formulated directly on the SO(3) manifold. For the attitude control, a state observer is constructed to recover the unmeasured states of flexible appendages, and a distributed backstepping-based scheme is employed to drive the group toward attitude agreement while attenuating structural vibrations. For the translational coordination, a sliding-mode-based method is adopted with an adaptive boundary layer and a radial basis function neural network. A Lyapunov analysis confirms the effectiveness of the overall controller in the assembly missions. Numerical tests together with air-bearing experimental results further demonstrate the feasibility and performance of the proposed approach.
Autonomous exploration in unknown environments requires accurate localization, efficient target selection, and reliable path planning. This work presents a multi-rover exploration framework that integrates fusion-based localization. The localization module fuses visual SLAM with periodic absolute pose corrections to suppress drift and maintain stable trajectory estimation. Candidate poses are scored by a weighted combination of path cost and expected information gain, dynamically adjusted to avoid redundant exploration among multiple rovers. Navigation paths are generated using an A*-based planner with multi-level safety maps and refined via triangle pruning to produce smooth and feasible trajectories. The framework is validated in a Webots-based lunar surface simulation with realistic rover and terrain models. Results demonstrate that the localization module effectively reduces trajectory errors, incorporating information gain accelerates single-rover exploration, and multi-rover coordination improves spatial coverage and reduces mission time.
Multi-layer satellite networks (MLSNs) have become a pivotal architecture for global communication coverage due to their enhanced capacity and spectral efficiency. However, existing studies primarily focus on constellation deployment and static cost optimization, overlooking dynamic operational overheads from cross-layer mobility management and routing control overhead. Addressing this gap, we pioneer the integration of network operation overhead into cost-benefit analysis from an MLSN networking perspective. Specifically, we establish a unified quantification methodology for both construction costs and operational expenditures, proposing a comprehensive cost-benefit evaluation framework for MLSNs. We formulate the MLSN topology and network operation design (MTNOD) problem to co-optimize topology parameters and operation strategies, simultaneously minimizing construction costs and operational overheads while maximizing flow-level capacity and transmission reliability. The Penalty Non-dominated Sorting Genetic Algorithm II (P-NSGA-II) algorithm efficiently generates robust Pareto frontiers, providing satellite operator with quantitative cost-performance trade-off benchmarks. Extensive simulations demonstrate the method's effectiveness in constructing MLSNs with low construction costs and minimal operation overhead.
On-orbit servicing covers essential tasks satellite maintenance, orbital assembly, and debris removal in modern space missions. A key prerequisite for such missions is the capability of a servicing spacecraft to inspect and characterize an unknown space object prior to any interaction. This paper presents a navigation and mapping algorithm that reconstructs the unknown object's shape using a factor graph-based Simultaneous Localization and Mapping (SLAM) method. The approach utilizes LiDAR point clouds with extracted FPFH (Fast Point Feature Histograms) descriptors as landmarks and performs batch optimization over a sliding window to enable real-time mapping. The proposed algorithm is designed for dynamic scenarios, specifically addressing the challenges posed by tumbling space objects. A dynamic SLAM formulation is developed by incorporating an uncertain parametric motion model to propagate the evolving map and construct a dynamic factor graph at the front-end. Kinematic factors are introduced to enable loop closures of previously observed features as they rotate and evolve with the tumbling object. These kinematic factors preserve the latest estimates of dynamic model parameters and connect consecutive landmark nodes in the factor graph to improve robustness and estimation accuracy. Numerical simulations demonstrate that, in static case, the estimated point cloud converges to the true map after a few iterations of batch factor graph optimizations. In dynamic cases, the algorithm effectively tracks map evolution, with mapping errors significantly reduced as the number of loop closures and feature associations increases. The proposed method is benchmarked against an ICP-SLAM pipeline with odometry and an EKF-SLAM approach, showing consistently lower trajectory and map errors at the cost of higher computation time. Overall, the findings highlight the potential of dynamic SLAM for equipping servicing spacecraft with advanced perception intelligence, enabling the reliable inspection and characterization of unknown targets in orbit.
Trajectory optimization has been widely used in dynamic soaring research to generate optimal flight patterns and analyze their performance. However, these solutions are typically derived for individual soaring cycles and do not directly translate into flight strategies. This study develops a systematic approach for deriving dynamic soaring strategies from optimal trajectory patterns. The approach identifies representative optimal patterns and evaluates the feasibility of non-powered dynamic soaring. Based on this assessment, appropriate patterns are selected to construct flight strategies. Non-powered patterns are adopted when sustained wind energy extraction is feasible, whereas thrust-augmented patterns are employed when propulsion support is required to maintain the flight cycle. The proposed approach is investigated in a wind environment with nocturnal low-level jets to demonstrate its applicability in strong low-altitude wind shear. Comparative analyses against dynamic soaring without nocturnal low-level jets and steady level flight without wind show that the derived strategies enable long-range dynamic soaring and improve propulsive energy efficiency.
Parabolic reflector antennas are widely employed in synthetic aperture radar (SAR) satellites for Earth observation due to their high gain and low sidelobe characteristics. Accurate estimation of antenna attitude is critical for space situational awareness (SSA) and strategic intelligence collection. This article proposes a joint estimation framework for the satellite-borne parabolic antenna attitude and the corresponding ground observation area by leveraging interferometric phase information extracted from inverse synthetic aperture radar (ISAR) imagery. The core innovation lies in exploiting the coplanarity of antenna edge scatterers to derive an explicit analytical relationship between interferometric phase differences and the antenna attitude parameters. This enables attitude estimation through a phase-based optimization model that eliminates the need for interferometric phase unwrapping. Building on this, a closed-form expression is derived that maps the attitude parameters directly to the geographic coordinates of the ground observation area. Furthermore, the impact of baseline configurations in bistatic ISAR systems on attitude estimation accuracy is analyzed, providing practical guidelines for optimal baseline deployment. Extensive simulations demonstrate the proposed method's high accuracy and robustness under challenging imaging conditions.
Current cooperative multi-task assignment models do not adequately capture electronic signal-guided imaging reconnaissance, where UAVs can execute tasks from feasible stand-off positions rather than by reaching exact target coordinates. This mechanism changes the evaluation of route length, task timing, energy consumption, and task feasibility. Accordingly, a heterogeneous cooperative multi-UAV reconnaissance task assignment model is formulated with actual flight distance as the optimization objective and payload- and maneuver-dependent energy consumption explicitly considered. Meanwhile, EIR–EOR precedence, task time windows, and heterogeneous task-resource compatibility substantially narrow the feasible solution space. To address this constrained search structure, an adaptive dual-group variable neighborhood search algorithm is proposed, combining diversification- and exploitation-oriented neighborhoods with a timing conflict repair strategy that restores EIR–EOR temporal feasibility. Evaluations across eight instances demonstrate that ADGVNS outperforms five baseline algorithms in diversity and convergence, particularly in large-scale scenarios (instance 6), reducing actual flight distances by up to 13.07% and computational costs by up to 82.64% compared to baseline algorithms.
Air-bearing satellite simulators are essential ground-based platforms for testing spacecraft guidance, navigation, and control technologies. To improve their precision and agility, cable-driven parallel robots offer a promising alternative to conventional pneumatic actuators, though they pose challenges in tension constraints and coordinated control. While optimization-based methods have shown promise in cable-driven parallel robot trajectory tracking and coordination, they suffer from scalability issues in highly redundant systems and are sensitive to motion uncertainties due to heavy reliance on the Jacobian matrix. This paper presents a coordinated control approach that combines a control-based positive tension strategy with an actor-critic framework. The method guarantees positive cable tensions without iterative optimization, enabling coordinated task-space trajectory tracking. Specifically, a saturation-like function is introduced to circumvent the computational burden and feasibility risks associated with iterative optimization, while contour error formulation is utilized to extend coordination from cable space to task space. The actor-critic architecture compensates for uncertainties such as Jacobian mapping errors in tension computation. The closed-loop system is rigorously proven to be semi-globally uniformly ultimately bounded based on the Lyapunov method. Finally, a cable-driven air-bearing satellite simulator is developed to address issues such as low control accuracy and poor directional control in pneumatic actuation. Experimental results on this platform validate the effectiveness of the proposed coordinated control approach.
This paper studies a multiplayer reach-avoid differential game in which a non-cooperative target moves along a fixed trajectory. The faster pursuers cooperate to escort the target moving from the start point to the goal point of the trajectory against the evaders, while the slower evaders aim to attack the target before the target reaches the goal point. The trajectory consists of two line segments known by all players. To handle the complexity of the multiplayer game, it is decomposed into many subgames with multiple pursuers and a single evader. When the target moves along the first segment, a pursuit strategy and the conditions are first proposed that guarantee the safety of the target against the evader regardless of the evasion strategy. Based on this, the pursuit strategy is then extended to two segments, and the conditions on states and parameters to ensure the safe arrival at the goal point throughout the trajectory are presented. Finally, such subgame outcomes are used to determine the pursuer-evader matchings and can guarantee a lower bound on the number of defeated evaders. Numerical and experimental results are presented to illustrate the theoretical results.
