In overflying Low Earth Orbit (LEO) satellite beam hopping (BH), the rapid variation in topology, intermittent visibility, dynamic traffic evolution, and the large joint beam and power action space collectively lead to significant challenges in beam scheduling. Traditional deep reinforcement learning (DRL)-based methods require exhaustive exploration of beam positions and power combinations, resulting in excessive training time and high computational complexity, which limits their practical deployment during LEO satellite overflight. Future LEO satellite networks demand high throughput to meet the needs of various applications, but the intelligent beam scheduling schemes for overflying satellites face high training complexity. These challenges become more critical for satellite-assisted IoT services, where massive low-rate terminals require wide-area and efficient access. To address this issue, this paper investigates the BH problem for overflying LEO satellite clusters and proposes a complexity-efficient DRL framework. To mitigate the excessive action space and the difficulty of acquiring channel state information during satellite overflight, a statistical pre-association mechanism is first introduced to eliminate infeasible satellite-cell pairs based on long-term traffic and visibility information. On this structured decision space, a constrained scheduler based on Multi-Agent Proximal Policy Optimization performs coordinated beam selection. Furthermore, fractional programming (FP) based power control is embedded into the learning loop to generate high-reward guidance signals, improving policy learning efficiency. Simulation results demonstrate that the proposed framework outperforms conventional DRL-based methods in terms of throughput and delay. Compared to traditional DRL-based methods, the average throughput is increased by 5-30%, and the training time is reduced by 50-60%.

In the maritime Internet of Things (MIoT), automatic identification system (AIS) devices serve as mobile sensing nodes at sea and rely on satellite-based reception and relaying for global ship situational awareness. However, their signals are susceptible to spoofing and deception, posing a potential threat to maritime security. Satellite constellation design enables the effective detection and localization of non-cooperative AIS signals by optimizing satellite orbits in targeted regions. Nevertheless, balancing positioning performance with deployment cost remains a central challenge in constellation design. This paper proposes an average positioning error (APE)-driven satellite constellation design scheme. First, a global AIS signal distribution model is constructed using real-world AIS data. The probabilities of different coverage multiplicities under dynamic network topologies are analyzed, and benchmark localization errors for various positioning regimes are established. Based on these insights, an analytical APE formula is derived to quantitatively evaluate the positioning performance of satellite constellations. Then, we model the constellation design as a multi-objective continuous optimization problem and propose a balanced adaptive constrained genetic algorithm (BACGA) to optimize the constellation configuration parameters. Simulation results show that under the minimum positioning error constraint, the proposed algorithm generates the optimal low-Earth orbit constellation configuration that meets accuracy requirements and achieves good coverage of key areas at low cost.

Traditional LLM-based vulnerability detection methods face challenges like hallucinations and high false positive rates. In order to overcome these constraints, we propose an innovative RAG-based method for inter-procedural vulnerability detection named IPVRAG. The main innovation design of our IPVRAG is its multi-level feature extraction strategy: during knowledge base construction, it not only extracts functional semantics and vulnerability causes but also stores pruned Data Flow Graph structures and semantic identifier information. Evaluated on a widely-used dataset, IPVRAG outperformed many of LLM-based baselines. It achieved optimal overall performance in inter-procedural vulnerability detection, particularly demonstrating superior precision-recall balance that effectively reduced false negatives.

This paper addresses the challenge of model mismatch in traditional time-of-arrival (TOA)-based localization methods for near-field or far-field source. We propose a unified TOA-based localization model that operates effectively in both scenarios. To estimate source direction, we introduce an algebraic closed-form solution, the two-step weighted least squares based on the modified polar representation (TSWLSMPR) method, which mitigates the issues associated with model mismatch. Additionally, we derive the Cramér-Rao Lower Bound (CRLB) as a benchmark to assess the performance of the proposed method. Simulation results demonstrate that the TSWLS-MPR method achieves the CRLB under varying levels of measurement noise, source distance, and source direction, confirming its accuracy and efficiency.

The integration of beam hopping (BH) technology into Low Earth Orbit (LEO) satellite communication systems has emerged as a critical strategy to enhance spectral efficiency and flexibility. However, conventional BH methods often exhibit insufficient robustness when confronted with uneven terrestrial traffic demands and dynamic jamming levels. This paper addresses this challenge by proposing a novel hybrid framework that synergizes statistical planning with multi-agent reinforcement learning (MARL) to achieve anti-jamming beam hopping for LEO satellites. The proposed framework decomposes the BH scheduling process into two phases: statistical planning and real-time adjustment. A portion of the beams are managed using a low-complexity potential game algorithm for statistical planning, ensuring stability and efficient resource allocation based on statistical traffic patterns and jamming conditions. The remaining beams are dynamically adjusted using a low information exchange Mean-Field Multi-Agent Proximal Policy Optimization (MFMAPPO) algorithm, enabling rapid adaptation to instantaneous jamming events and fluctuating user demands. This framework reduces the number of beams considered during training, thereby simplifying the action space. As a result, the system achieves quicker convergence and offers enhanced robustness, especially in environments where some information may be missing. Simulation results demonstrate that the proposed method significantly enhances throughput, improves robustness against jamming, and reduces training time compared to other baselines. The integration of statistical planning and mean-field MARL effectively balances long-term efficiency with real-time adaptability, achieving high-quality communication in low Earth orbit satellite coverage areas under high dynamic environments.

Fluid Antenna (FA) systems hold significant potential for enhancing physical layer security (PLS) by dynamically adjusting the positions of transmit antennas to suppress information leakage to eavesdroppers. However, the joint optimization of secure beamforming and FA positions is very challenging and remains unsolved, given the mutually coupled, non-convex and NP-hard nature of the problem. In this paper, we investigate the FA-assisted multi-user multiple-input single-output (MU-MISO) system for maximizing the downlink secrecy rate. First of all, we propose an alternating optimization (AO) framework to decouple the problem. For efficient FA position optimization, we introduce a low-sampling successive selection and successive convex approximation (L3S-SCA) method, which first selects a proper port in discrete space and subsequently refines the FA positions via continuous optimization. For secure beamforming, we reformulate the problem as an unconstrained optimization on Riemannian manifold, eliminating the errors from relaxing per-antenna power constraints (PAPC). We design the necessary Riemannian tools and propose a Limited-memory Riemannian Broyden-Fletcher-Goldfarb-Shanno (LRBFGS) method with low computational complexity. Comprehensive convergence and complexity analyses are conducted, and simulation results demonstrate the advantages of FA-assisted secure beamforming, as well as the superiority of our proposed algorithms in terms of performance and complexity.

The multibeam satellite system (MBSS) has great potential for mobile communications in 5G era due to its superiority in terms of extensive coverage, large capacity and real-time service. In order to integrate the resource allocation in multiple dimensions and maximize the system capacity of the MBSS, the scenario of a sparse network with dense users is selected to investigate the resource allocation method in time dimension, frequency dimension, space dimension and power dimension. We first propose a multilevel clustering algorithm and a cross-cluster grouping algorithm to realize the beam scheduling, by which the interference is reduced in time dimension and space dimension. Based on the beam scheduling scheme, we further explore the relationship between the system capacity and the resource allocation in frequency dimension and power dimension, where a joint power allocation and subchannel selection algorithm is proposed to optimize the spectral efficiency. Our simulation results show that the proposed multiple-dimension resource allocation method is superior to the existing methods in system capacity and convergence, which is not only applicable for the resource allocation in the MBSS but also provides an efficient approach to solve the coupling resource allocation problem.

In data relay satellite (DRS) systems, the performance of tasks scheduling is influenced by the variation of task and resources, which degrades the processing capacity of relay satellites. Considering this problem, we investigate the dynamic scheduling in the application of DRS. To achieve the efficient resource utilization and reliable data transfer, the strategies of task preemptive switching and decomposition are designed. Based on the initial scheme, we construct a dynamic scheduling model with multiple objectives, including maximizing the total weight of scheduled tasks, minimizing the change of scheduling scheme and minimizing the number of decomposed subtasks. Meanwhile, a preemptive dynamic scheduling algorithm (PDSA) is designed to solve the proposed model. Explicitly, our simulation results show that PDSA is superior to the whole rescheduling algorithm (WRA) in quantities of completed tasks, rescheduling rate of scheme and processing time, which can efficiently improve the performance of dynamic scheduling in DRS systems.