With the increasing demand for secure connectivity in non-terrestrial networks, physical layer security (PLS) in Low Earth Orbit (LEO) satellite communications has attracted growing attention. This paper investigates PLS in the presence of uncertain eavesdropper locations, where satellite beamforming is used to enhance signal strength toward legitimate users while suppressing emissions toward potential eavesdropper regions. A joint optimization framework is developed for satellite beamforming and UAV-assisted artificial noise (AN) design to maximize the secrecy capacity. The resulting non-convex problem is addressed using alternating optimization (AO), semidefinite relaxation (SDR), and successive convex approximation (SCA). Numerical results confirm that the proposed method achieves superior secrecy performance compared to baseline approaches. Additionally, we study the impact of region size, relative location, and satellite mobility on the achievable secrecy rate.
This paper proposes an end-to-end autonomous driving method for intelligent buses integrating vehicle dynamics. By simultaneously using the motion state and dynamic information of the intelligent buses as inputs to the end-to-end network, the convergence speed of the network training is accelerated, and the decision accuracy is improved. Specifically, a dynamic model is established that can accurately represent the changes in the dynamic characteristics of intelligent buses. Moreover, an end-to-end autonomous driving network incorporating spatiotemporal features and attention mechanisms is constructed to determine lateral and longitudinal control commands based on images, motion states, and dynamic information. On this basis, the PreScan simulation platform is utilized to obtain driving scenarios that collect images and driving datasets. Finally, a simulation test is conducted to demonstrate the feasibility and effectiveness of the proposed method. The results show that the proposed methods can reduce training loss by 48.26% and testing loss by 39.6% compared with the method using the end-to-end model without integrating vehicle dynamics.
The integration of IEEE 802.11 Wi-Fi networks in dense deployments faces significant challenges related to resource management, particularly with the introduction of advanced multilink capabilities in Wi-Fi 7. Effective coordination Access Points (APs) is essential for optimizing communication reliability and maximizing overall spectrum efficiency. This research introduces a Graph-Driven Min-Max Link Scheduling (GMMLS) approach, which models traffic load distribution among Multilink-Enabled Devices (MLDs) and facilitates optimal link allocation through enhanced AP coordination. Our extensive evaluations using the NS-3 simulator demonstrate that managing the new multilink operation feature with GMMLS enhances Service Level Agreement (SLA) compliance by up to 88% compared to existing scheduling techniques, making it a highly promising solution for denser environments.
To address the challenge of anti-jamming transmission in wireless communication systems under complex interference environments, this paper proposes a movable antennas systems anti-jamming transmission scheme based on deep unfolding networks. Although existing research has enhanced the diversity of anti-jamming strategies through mobile antenna technology, its optimization algorithms are complex, system overhead is large, and traditional deep learning schemes face problems such as strong data dependence and poor interpretability. Therefore, this paper innovatively introduces deep unfolding networks into the design of movable antennas systems, achieving efficient optimization through the fusion of algorithm unfolding and neural networks. Specifically, for the multi-jammer confrontation scenario, the non-convex joint optimization problem is decoupled into an alternating optimization problem of the receiving beamforming vector and antenna position, where the beamforming is solved in closed form through the generalized Rayleigh quotient. Further, by mapping the iterative optimization steps to the hidden layers of the neural network, a lightweight deep unfolding framework is constructed, significantly reducing the computational complexity of traditional iterative algorithms. Simulation results verify the effectiveness and reliability of the proposed scheme in real-time anti-jamming transmission under jamming environments.
Mobile edge computing (MEC) enables efficient computation offloading for mission-critical applications in resource-constrained vehicles, while reconfigurable intelligent surface (RIS) help address connectivity challenges for vehicles in urban environments with severe signal blockages. Non-orthogonal multiple access (NOMA) is an appealing technique that improves spectral efficiency while mitigating multi-user interference. This work proposes the RIS-assisted NOMA-MEC in vehicular networks, considering dynamic challenges such as heterogeneous vehicle processing capability, time-varying channel from high-mobility and dynamic task workloads. We formulate a system latency minimization problem by jointly optimizing the task offloading ratio, edge server resource allocation and RIS passive beamforming, while satisfying the task deadline and Signal to Interference plus Noise Ratio (SINR) requirements. To overcome the limitations of conventional optimization methods in such dynamic environments, we propose a soft actor critic (SAC)-based deep reinforcement learning (DRL) framework, which dynamically adapts to real-time channel state information (CSI), task workload and vehicle processing capability of all vehicles. Simulation results demonstrate that our approach achieves lower latency performance compared with the Deep Deterministic Policy Gradient (DDPG) baselines. Moreover, the proposed SAC method exhibits robustness and adaptivity to various levels of uncertainty in the CSI.
The finite-field multiple-access (FFMA) technique reverses the order of channel coding and multiplexing, thereby enhancing the error performance of multiuser transmission, which is particularly promising for next-generation communication systems. In an FFMA system, element-pairs (EPs) serve as virtual resources. Previous studies on FFMA systems focused on assigning a single EP to each bit. In this paper, we investigate the case where multiple EPs are assigned to each bit, with the EPs selected from the orthogonal EP set Psi(o,B). We propose a uniform random EP allocation scheme (UR-EPA), in which each bit is assigned an identical number of EPs with a randomized assignment feature. We prove that the orthogonal EPs assigned by the UR-EPA scheme satisfy the unique sum-pattern mapping (USPM) constraint. Under this scheme, the orthogonal EP code Psi(o,B) transforms into a single codeword EP code (S-CWEP), effectively forming an interleave-division multiple-access system in the finite-field domain (FF-IDMA). Furthermore, we propose a Turbo bifurcated minimum distance (Turbo-BMD) decoding algorithm for decoding multiuser signals. Simulation results show that, with 100 users and an average BER of 10(-5), the FF-IDMA system achieves approximately 1.2 dB of coding gain compared to the classical IDMA system.
This paper considers channel estimation for extremely large-scale intelligent reflecting surface (XL-IRS)-assisted terahertz (THz) communication systems. Specifically, an XL-IRS is deployed close to users (UEs) to enhance communication performance between the base station (BS) and UE. With its large aperture, the XL-IRS has a Rayleigh distance of tens of meters. Therefore, the users are likely located in the near-field region of the XL-IRS, while the BS is in its far-field region. Consequently, a spherical wavefront propagation model should be considered to characterize the propagation property between the XL-IRS and the UE, while the planar wavefront propagation model is utilized in the BS-IRS link. By leveraging Khatri-Rao product and Kronecker product properties, we rephrase the channel estimation problem. In addition, we construct an orthogonal dictionary, which essentially modifies the well-known Discrete Fourier Transform (DFT) matrix. We further find that the considered channel can be block-sparsely represented by this dictionary. Hence, the channel estimation can be converted into a block sparse recovery problem, which can be efficiently solved by several off-the-shelf methods. The simulation results show that our proposed method achieves better estimation performance than the conventional polar-domain-based method.
In urban environments, numerous IoT devices rely on accurate radio map information to enable critical applications such as localization, trajectory planning, and communication. The effectiveness of these applications hinges on the availability of high-fidelity and spatially comprehensive radio maps. To address this, we propose an optimal path planning approach with distance constraints for efficient sampling in urban radio map reconstruction. Unlike traditional random sampling strategies, our method restricts sampling points to lie along a feasible road network and imposes a fixed distance constraint on the sampling path. We design a structure-aware genetic algorithm (GA_s) to optimize the path for mobile sampling, incorporating an unsupervised population evaluation metric to assess the fitness of candidate solutions. Experimental results show that, under equal sampling distance conditions, GA_s achieves a Root Mean Square Error (RMSE) of 0.0626 in radio map recovery—outperforming random sampling (0.0796), A* algorithm (0.0748), and a conventional genetic algorithm (GA_c). These results demonstrate the effectiveness of our method in enabling efficient and accurate radio map reconstruction for mobile sampling platforms in real-world urban settings.
The vehicle-road-cloud continuum represents a promising frontier for cyber-physical systems, particularly in the realization of safety-critical V2X services. While containerization and orchestration frameworks such as Kubernetes offer scalable resource management, their dependability assumptions diverge significantly from the stringent safety guarantees required in the automotive domain. This paper presents a dependability-aware coordinator, inspired by automotive safety principles, designed for deployment at roadside infrastructure. We implement a prototype atop Kubernetes and deploy it across multiple urban intersections. The coordinator continuously monitors V2X warning pipelines, detects faults, and dynamically reallocates workloads to maintain service continuity. Through a connected automatic emergency braking use case based on infrastructure-side perception, we demonstrate how mainstream IT technologies can be extended and adapted to fulfill automotive-grade dependability requirements. This work bridges the methodological gap between automotive safety assurance and modern cloud-edge orchestration, advancing the safe deployment of V2X services.
Radio Map Estimation (RME) plays a crucial role in wireless network planning and optimization, enabling efficient resource allocation and interference management. Unlike traditional methods that rely on transmitter locations, sampling point-based approaches offer greater practical benefits by leveraging real-world measurements. In this paper, we argue that RME should be treated not merely as a generative task but rather as a style transfer problem, where the goal is to transform sparse sampling point distributions into high-fidelity radio maps. To this end, we propose RadioPix2pix, a Generative Adversarial Network GAN-based model with physical awareness that effectively learns the mapping between sparse measurements and dense radio maps. Experimental results demonstrate that our method significantly outperforms deep learning baselines, achieving superior performance even at an extremely low sampling rate of 0.15%. This work provides a novel perspective on RME and advances the state-of-the-art in data-driven radio map reconstruction.
To address the challenges of spectral leakage and self-interference in filter bank multicarrier with quadrature amplitude modulation (FBMC/QAM) system under spectrally-constrained low-altitude economy (LAE) communication scenarios, a prototype filter design scheme is proposed. This scheme relaxes the orthogonality conditions between the two prototype filters to formulate a constrained optimization problem, which aims to minimize self-interference power subject to time and frequency localization constraints. A prototype filter with complex frequency-domain coefficients is derived for odd-numbered subcarrier symbols by solving the problem. Simulation results show that the FBMC/QAM system with the proposed filter achieves low out-of-band (OOB) radiation, and at the same time maintains good bit error rate (BER) performance, providing a practical solution for efficient deployment in LAE networks.
Most existing deep learning-based path loss (PL) models rely on measurements in a given frequency range and specific scenarios, making it difficult to balance accuracy and generalization. The emergence of pre-trained language model (PLM) provides a solution to this challenge. In this paper, we integrate prior knowledge, engineering data, and environmental features into PLM for multi-modal feature extraction and fusion, and use the Low-Rank Adaptation (LoRA) for lightweight fine-tuning to achieve cross-modal knowledge transfer. Moreover, in our model, a residual structure is introduced to bridge the gap between the statistical model and the measurement data. The experimental results show that the proposed model achieves a prediction error [root mean square error (RMSE)] of 3.5 +/- 0.2 dB compared to the 7.1 +/- 1.5 dB for 3GPP statistical model prediction, and the pearson correlation coefficient of 0.88. When extrapolated to new scenarios, the proposed model has excellent few-shot performance.
Low Earth orbit (LEO) satellite constellations play a pivotal role in sixth-generation (6G) wireless networks by providing global coverage, massive connections, and huge capacity. In this paper, we present a novel LEO satellite constellation communication framework, where a reconfigurable intelligent surface-mounted unmanned aerial vehicle (RIS-UAV) is deployed to improve the communication quality of multiple terrestrial user equipments (UEs) under the condition of long distance between satellite and ground. To reduce the overhead for channel state information (CSI) acquisition with multiple-satellite cooperation, statistical CSI (sCSI) is utilized in the system. In such a situation, we first derive an approximated but exact expression for ergodic rate of each UE. Then, we aim to maximize the minimum approximated UE ergodic rate by the proposed alternating optimization (AO)-based algorithm that jointly optimizes LEO satellite beamforming and RIS phase shift. Finally, extensive simulations are conducted to demonstrate the superiority of the proposed algorithm in terms of spectrum efficiency over baseline algorithms.
This paper investigates robust transmit beamforming based on statistical channel state information (CSI), against satellite-to-terrestrial user terminal (UT) interference arising from spectrum sharing in the integrated terrestrial and satellite network. First, we develop an integral-form interference model free of shared CSI to characterize the interference from satellite to terrestrial UTs. Then, we propose a robust interference-avoidance transmit beamforming scheme under the interference threshold and power budget. We derive a closed-form solution based on the minimum mean square error criterion and apply a bisection method to satisfy interference thresholds. Furthermore, we introduce a base station position-aided approximation scheme to eliminate the complex integral calculations. Numerical simulations validate the proposed schemes.
This paper proposes the integration of the Multiple Signal Classification (MUSIC) algorithm with Affine Frequency Division Multiplexing (AFDM)-based Integrated Sensing and Communications (ISAC) system, enabling high-accuracy joint range-velocity estimation for multiple targets. To address the high computational complexity of conventional MUSIC, we propose a low-complexity iterative MUSIC (LCI-MUSIC) variant that employs cyclic spectral peak search instead of uniform grid scanning. This innovation achieves an exponential reduction in computational complexity while maintaining detection accuracy. In addition, the LCI-MUSIC enables trade-offs between complexity and estimation accuracy. Simulations demonstrate that compared to conventional MUSIC, the LCI-MUSIC algorithm achieves higher estimation accuracy at the same computational complexity.
In the high-speed railway (HSR) scenario, handover (HO) reliability is constrained by the high-speed movement and weak coverage at cell edges, posing a threat to the "always online" transmission requirement. To address this, we propose a dynamic beamforming scheme to mitigate HO failures and improve data rates. First, a beamforming-based HO model is established, quantifying the impact of beam direction on data rate and HO reliability. Based on this, an optimization problem is formulated, aiming to enhance data rate, reduce HO failure probability, and minimize beam adjustment overhead. Subsequently, a dynamic beam adjustment algorithm based on deep reinforcement learning is designed, leveraging its real-time decision-making capability to adaptively optimize the beam direction under rapidly changing channel conditions. Simulation results demonstrate that the proposed scheme requires only 23% of the beam adjustment overhead of the ideal real-time precise beamforming, while achieves nearly identical HO performance and 98.9% of its data rate.
Ambient backscatter devices have been demonstrated to have significant potential in a variety of IoT applications. These devices are notable for their low-cost and low power consumption, which facilitates their massive deployment in numerous scenarios. However, due to the extensive deployment as well as the constraints of low power and cost, traditional random access protocols are not well-suited to such backscatter communication scenarios. In this article, a demodulation strategy for two packets colliding is presented, which was realised by analysing the collision constellation diagram and the meanings of collision clusters with a training sequence. Subsequently, this strategy is used to design a random access protocol for this large-scale access scenario. In addition, we also analyse the performance of the proposed protocol using a bipartite graph and design an optimization problem to gain the suboptimal probability distribution of the number of packet copies to further improve the access performance. The final simulation results imply that the performance of the proposed protocol is significantly enhanced in comparison to traditional protocols.
In-vehicle networks (IVNs) are rapidly evolving to support increasingly complex automotive applications, demanding higher bandwidth and deterministic timing bounds. Time-Sensitive Networking (TSN) has emerged as a promising Ethernet-based technology that addresses these stringent requirements. However, evaluating TSN-based IVN strategies remains a challenge due to the lack of standardized benchmarks and simulation tools. This paper introduces INSIM, a modular simulation platform specifically designed for TSN-based IVNs, providing an intuitive graphical interface, an extensible plug-in architecture, and integrated benchmarking features. INSIM integrates analytical performance models and discrete-event simulations (as plug-ins), enhancing the workflow for engineers by refining topology design, adjusting parameters, conducting simulations, and assessing performance, while providing researchers with a flexible platform to plug in, analyze, and compare custom network resource managers or analytical performance models.
Dynamic wireless channel modeling is essential for future communication system design. However, existing methods struggle to capture the long-term non-stationary behavior of real-world channels driven by high mobility, dense deployment, and complex environments in the sixth generation (6G) scenarios. As system-level design depends on channel statistics rather than exact realizations, statistically consistent channel generation offers a more appropriate modeling strategy. In this work, We propose a deep learning-based hybrid framework that shifts the modeling goal from precise point-wise prediction to generating synthetic channel sequences that statistically replicate long-term non-stationary dynamics. Instead of forecasting exact future states, the model reproduces key statistical features such as the evolution of power delay profiles (PDPs), root mean square (RMS) delay spread and wide-sense stationary (WSS) regions. Experimental results demonstrate that the generated sequences closely match reference data, supporting scalable data generation for digital twins and system-level simulations under realistic dynamic conditions.
Millimeter-wave (mmWave) radar has emerged as a promising sensing technology for various applications due to its capabilities of all-weather operation and direct velocity measurement. However, existing mmWave radar schemes exhibit significant shape reconstruction errors when the target is partially occluded by obstacles. To address this issue, we propose a wireless amodal sensing paradigm that supports the shape reconstruction of the occluded target using a single mmWave radar. The basic idea is to first extract the features from the mmWave point cloud of the occluded target, and then leverage pre-trained generative models to complete the whole shape based on these features. New challenges have arisen that mmWave radar point clouds are inherently sparse, containing measurement noise that reduces the accuracy of feature extraction, thus resulting in degraded shape reconstruction quality. To tackle these challenges, we propose WASNet that incorporates a radar cross section (RCS)-enhanced geometric feature extraction module to suppress measurement noise and a Vision-Language Model (VLM)-based semantic injection module to enhance the shape reconstruction accuracy by extracting semantic features. Experiments demonstrate the effectiveness and robustness of the proposed scheme, which achieves a 70.5% reduction in point cloud reconstruction error compared with baselines.
