Recently, semantic communication has emerged as a promising solution for the efficient transmission of point cloud data. However, effectively extracting robust semantic features and achieving precise recovery under channel impairments remain formidable challenges. Recognizing the critical role of high-order correlations in compact representation and robust recovery, we propose a High-order Feature-guided Semantic Communication System (HFSCS). At the transmitter, a high-order structured semantic encoder integrating hypergraph convolution with lightweight squeeze-and-excitation attention mechanism is devised to extract and refine high-order features. These features are subsequently projected into a compact semantic vector via a feature compression and mapping module, thereby significantly reducing bandwidth overhead. At the receiver, we design a noise-resilient semantic reconstruction architecture comprising a semantic space recovery decoder and a prior-guided refinement module. Treating the received semantics as structural priors, it performs coarse-to-fine denoising and multi-level progressive fusion within a high-dimensional space, accurately recovering fine-grained geometric details. Extensive experiments across AWGN, Rayleigh, Rician, and 3GPP channels demonstrate that HFSCS significantly outperforms traditional codecs and the state-of-the-art semantic communication baseline in reconstruction accuracy. Notably, it exhibits strong robustness against low SNR, low Channel Bandwidth Ratio (CBR), and training-test SNR mismatch. Furthermore, compared to the baseline semantic approach, HFSCS improves codec efficiency by 81.82% while reducing FLOPs by 52.02%.

Underwater wireless optical communication (UWOC) channels are highly sensitive to water-condition variations, bubble disturbance, and transmitter-receiver misalignment, which makes timely channel-state perception challenging for adaptive transmission. This paper proposes a vision-based channel-state perception and parameter-efficient adaptation method (PEFT) for dynamic UWOC channels. The proposed method uses underwater environmental images to extract channel-related visual cues with a DINOv2-small backbone. During source-condition training, turbidity, bubble disturbance, and link deflection angle are used as auxiliary supervision to regularize the learned representation toward propagation-related factors. During target-condition adaptation, the visual backbone is frozen, and only lightweight low-rank adaptation modules and task-specific heads are updated. In addition, an uncertainty-guided weighting strategy is introduced to emphasize target samples that are less familiar to the source model. Experiments on a laboratory-collected UWOC dataset show that the proposed method achieves better target-condition channel-state prediction than source transfer, target-only training, full fine-tuning, and unlabeled adaptation baselines, while requiring substantially fewer trainable parameters. These results demonstrate the potential of parameter-efficient visual adaptation for low-cost cross-condition UWOC channel perception.

This work investigates transmission schemes for a Gaussian diamond network in which the relays lack knowledge of the source codebook. A novel simultaneous amplify- and compress-forward (SAC) scheme is proposed in which each relay splits its transmission power for amplifying the received signal and forwarding a compressed version of its observation simultaneously. To address practical implementation constraints and asymmetric channel conditions, two simplified variants are further introduced, namely the asymmetric SAC (ASAC) and hybrid amplify-/compress-forward (HAC) schemes. In the ASAC scheme, one relay operates in the traditional amplify-forward (AF) mode, while the other simultaneously operates in both AF and compress-forward (CF) modes. In the HAC scheme, the two relays adopt a different scheme, with one performing pure AF and the other pure CF. The achievable rate of the SAC scheme is characterized through theoretical analysis, and an optimization problem is formulated to maximize the achievable rate. In addition, the closed-form expressions for the achievable rates of ASAC and HAC are derived to enable efficient evaluation. The simulation results demonstrate that the proposed SAC scheme achieves notable rate improvements by jointly leveraging the benefits of AF and CF. Meanwhile, the ASAC and HAC schemes offer low-complexity alternatives suitable for practical implementations.

The ability to ensure covert unmanned aerial vehicle (UAV) communications is imperative in critical missions such as military surveillance and emergency response. In this paper, a multi-UAV covert communication system with informed jammers is investigated. To increase ground wardens' detection uncertainty, we propose a joint dynamic scheduling and sensing jamming (DSSJ) scheme. Unlike existing approaches with fixed UAV roles and non-informed jamming, DSSJ dynamically schedules UAVs across adjacent time slots (TSs), while the jamming UAV performs sensing-based informed jamming per TS. Closed-form expressions are derived for the covert rate and minimum detection error probability (MDEP) under the worst-case scenario with optimal warden detection. An optimization problem is formulated to maximize the normalized weighted sum of covert rate and MDEP, subject to multiple constraints, including scheduling, sensing ratio, and other key factors. To solve this mixed-integer non-convex problem, we design a double deep Q-network (DDQN)-DSSJ algorithm, integrating DSSJ within a deep reinforcement learning framework, accelerated by experience replay and dynamic exploration, achieving real-time covert decision-making with polynomial complexity. Simulations demonstrate that DDQN-DSSJ achieves 25% faster convergence, enhanced stability, and superior covertness compared to proximal policy optimization and deep Q-network. Additionally, DDQN-DSSJ improves the covert rate by over 4x and MDEP by up to 28.3%, outperforming state-of-the-art schemes.

Low-altitude Unmanned Aerial Vehicle (UAV) networks rely on semantic segmentation as a core perceptual capability enabling distributed sensing-communication-control co-design across heterogeneous agents. However, segmentation foundation models deteriorate quickly under weather, lighting, and viewpoint drift. Resource-limited UAV cannot run gradient-based test-time adaptation, while resource-massive UAV adapt independently, wasting shared experience. To address these challenges, we propose AdaptFly, a prompt-guided test-time adaptation framework that adjusts segmentation models without weight updates. AdaptFly features two complementary adaptation modes. For resource-limited UAV, it employs lightweight token-prompt retrieval from a shared global memory. For resource-massive UAV, it uses gradient-free sparse visual prompt optimization via Covariance Matrix Adaptation Evolution Strategy, with offline prompt distillation for knowledge sharing. An activation-statistic detector triggers adaptation, while cross-UAV knowledge pool consolidates prompt knowledge and enables fleet-wide collaboration with negligible bandwidth overhead. Extensive experiments on UAVid and VDD benchmarks, along with real-world UAV deployments under diverse weather conditions, demonstrate that AdaptFly significantly improves segmentation accuracy and robustness over static models and state-of-the-art TTA baselines. The results highlight a practical path to resilient, communication-efficient perception in the emerging low-altitude economy.

In complex electromagnetic environments, the scale effect of Unmanned Aerial Vehicle (UAV) swarm presents significant potential for enhancing cooperative effectiveness. However, the accuracy of Time Difference of Arrival (TDOA)-based localization for non-cooperative emitters using UAV swarm is significantly affected by the coupling of multi-source errors, which mainly include UAV position error (UPE), clock synchronization error (CSE), and TDOA measurement error (TME). To address the challenges of evaluating cooperative effectiveness under multi-source errors coupling and balancing localization accuracy with computational efficiency, a cooperative utility of information (CUoI) optimization approach is proposed.First, a TDOA observation uncertainty model is constructed by integrating multi-source errors. Then, the information gain of target position estimation is derived to build the CUoI evaluation model. Next, the characteristic of Dueling Deep Q-Network (Dueling DQN) that decouples state value from action advantage is leveraged, enabling precise evaluation of the potential benefits of different hyperparameter adjustment strategies. This characteristic facilitates adaptive tuning of key hyperparameters in Particle Swarm Optimization (PSO). Finally, a dynamic PSO framework based on Dueling DQN is proposed to effectively balance localization accuracy and computational efficiency. Numerical experiments demonstrate that the proposed algorithm achieves reductions in average localization RMSE of 19.1%, 6.0%, and 1.4%, respectively, compared to Semidefinite Relaxation-TDOA (SDR-TDOA), Grey Wolf Optimizer (GWO), and Multi-swarm Discrete Quantum-inspired Particle Swarm Optimization with Adaptive Simulated Annealing (MDQPSO-ASA).

Recent advances in mobile edge computing (MEC) have transformed unmanned aerial vehicles (UAVs) into intelligent platforms capable of distributed sensing, computing, and communication. However, coordinating multiple UAVs presents significant challenges in task assignment, path planning, and resource allocation. To address these issues, we propose a joint task assignment and resource allocation optimization (TARO) approach for efficient target point (TP) visit in multi-UAV MEC systems. Specifically, considering resilience in this work, each TP may require multiple visits, while each UAV is restricted to visiting any given TP only once, leading to non-overlapping visit requirements. We formulate an optimization model as a mixed-integer nonlinear programming (MINLP) to minimize the maximum task completion time among all UAVs. To solve the problem efficiently, we adopt a Benders-inspired decomposition framework with infeasibility elimination, which separates the original problem into a routing master problem and two continuous subproblems. In addition, efficient solution approaches (e.g., closed-form solution) are developed to solve them respectively. Simulation results show that TARO significantly reduces latency and enforces temporally separated repeated visits to the same TP under various mission configurations and bandwidth constraints, thereby improving resilient mission performance.

Real-time speech communication over wireless networks remains challenging, as conventional channel protection mechanisms cannot effectively counter packet loss under stringent bandwidth and latency constraints. Semantic communication has emerged as a promising paradigm for enhancing the robustness of speech transmission by means of joint source channel coding (JSCC). However, its cross-layer design hinders practical deployment due to the incompatibility with existing digital communication systems. To address this, we perform JSCC over the network layer to combat packet loss and support real deployment. Inspired by the generative latent modeling, we propose Glaris, a generative latent-prior-based resilient speech semantic communication framework that performs resilient transform coding in the generative latent space. Generative latent priors enable high-quality packet loss concealment (PLC) at the receiver side, well-balancing semantic consistency and reconstruction fidelity. Additionally, an integrated error resilience mechanism is designed to mitigate the error propagation and improve the effectiveness of PLC. Compared with traditional packet-level forward error correction (FEC) strategies, our new method achieves enhanced robustness over dynamic wireless networks while reducing redundancy overhead significantly. Experimental results on the LibriSpeech dataset demonstrate that Glaris consistently outperforms existing error-resilient codecs, achieving JSCC-level robustness while maintaining seamless compatibility with existing systems, and it also strikes a favorable balance between transmission efficiency and error resilience.

This paper proposes a digital twin network (DTN)-driven framework for multi-unmanned aerial vehicle (UAV) covert communication with informed jammers. To maximize warden detection uncertainty, we design a dynamic scheduling and sensing jamming (DSSJ) scheme that dynamically schedules UAV roles (communication/jamming/backup) across adjacent time slots (TSs), while the jamming UAV employs sensing-based informed jamming per TS. Leveraging DTN’s real-time digital-physical synchronization, closed-form expressions for the user’s covert rate and warden’s minimum detection error probability (MDEP) are derived under worst-case detection. A DTN-driven double deep Q-network (DDQN)-DSSJ algorithm solves the normalized weighted-sum maximization problem under covertness, sensing ratio, power, and speed constraints. Simulations demonstrate that DTN-driven DDQN-DSSJ achieves 25% faster convergence, enhanced stability, and superior covertness compared to proximal policy optimization (PPO) and deep Q-network (DQN).

Data compression technology is able to reduce data size, which can be applied to lower the cost of task offloading in mobile edge computing (MEC). This article addresses the practical challenges for robust trajectory and scheduling optimization based on data compression in the uncrewed aerial vehicle (UAV)-assisted MEC, aiming to minimize the sum energy cost of terminal users while maintaining robust performance during UAV flight. Considering the nonconvexity of the problem and the dynamic nature of the scenario, the optimization problem is reformulated as a Markov decision process (MDP). Then, a randomized ensembled double Q-learning (REDQ) algorithm is adopted to solve the issue. The algorithm allows for higher feasible update-to-data ratio, enabling more effective learning from observed data. The simulation results show that the proposed scheme effectively reduces the energy consumption while ensuring flight robustness. Compared to the PPO and A2C algorithms, energy consumption is reduced by approximately 21.9% and 35.4%, respectively. This method demonstrates significant advantages in complex environments and holds great potential for practical applications.

Time-varying piecewise smooth data recovery problem exists extensively in computer vision, image processing, environment monitoring, etc. In recent years, the emerging field of graph signal processing (GSP) provides a new way to solve this problem, deriving the graph signal matrix completion (GSMC) which incorporates the correlation among data entries. The model-based methods of GSMC are more interpretable, but their reconstruction quality is still not satisfactory, especially when observations are sparse. In this paper, we propose a new matrix completion method to solve the time-varying data recovery problem. By jointly exploiting the graph difference operator and the time difference operator to capture the spatio-temporal correlation of the data, we obtain a method based on low-rank and piecewise-differential smoothness (LRPDS). The proposed method achieves high recovery accuracy by jointly exploiting low rank property, the piece- wise smoothness and differential smoothness of graph signals. Numerical results on three real-world datasets demonstrate that our scheme has better reconstruction performance compared with existing model-based matrix completion approaches.

Skeleton-based action recognition has broad application prospects in many fields such as virtual reality. Currently, the most popular way is to employ Graph Convolutional Networks (GCNs) or Hypergraph Convolutional Networks (HGCNs) for this task. However, GCN-based methods may heavily rely on the physical connectivity relationship between joints while lack the capture of higher-order information about interactions among distant joints, and HGCN-based methods usually introduce unnecessary noise when capturing low-order information of skeleton structures with simple topology. Besides, the current methods do not deal well with redundant frames and confusing frames. These limitations hinder the improvement of recognition accuracy. In this paper, we propose a novel network, called Hyper-Net, which combines multi-order representation learning and frame optimization learning for skeleton-based action recognition. Specifically, the proposed Hyper-Net contains Temporal-Channel Aggregation Graph Convolution (TCA-GC), Spatial-Temporal Aggregation Hypergraph Convolution (STA-HC) and Frame Optimization Learning (F-OL) modules. The TCA-GC aggregates low-order and local information from simple joint and bone topologies across different temporal and channel dimensions. The STA-HC captures high-order and global information from complex motion streams as well as solving the problem of spatial-temporal weight imbalance. The F-OL can adaptively extract key frames and distinguish confusing frames, thus improving the ability of the network to recognize confusing actions. A large number of experiments are conducted on the NTU RGB+D, NTU RGB+D 120 and NW-UCLA datasets for action recognition task. Experimental results demonstrate the superiority and effectiveness of the proposed network.

By fully exploring the edge computing "supply-demand" relationship between the mobile-edge computing (MEC) servers and the differentiated application requests, the computing pricing (i.e., "supply") and allocating (i.e., "demand") can be coordinated well for the practical network consisting of heterogeneous users and MEC operator. In this article, the fair-aware computing pricing, beneficial offloading (i.e., obtaining positive utility) and local computing adjustment are jointly discussed under a pricing-enabled MEC. By considering heterogeneous application requests, fair service demand and limited computing provisioning, a multiobjective composite utility optimization is developed to maximize the user utility and the MEC operator profit simultaneously. Therein, the fair service condition is proposed, under which each user can experience a similar chance to obtain beneficial offloading. In order to solve the goal problem with undetermined objective function and conditions, a fair service enabled pricing and allocating algorithm (FS_PAA) with extremely low complexity is proposed by exploiting classification discussion method and convex optimization. Our FS_PAA reveals the explicit relationship between the optimal offloading decision and computing pricing, and the explicit relationship between the optimal computing pricing and the maximum computing provisioning, which helps to provide an effective reference for practical edge computing deployment. Simulation results show that our FS_PAA can 1) ensure fair offloading services for practical differentiated requests; 2) provide green offloading service for more users; 3) greatly improve the utilization of edge computing resource.

Hypergraphs can accurately capture complex higher-order relationships,but it is challenging to identify their impor-tant nodes.In this paper,an improved PageRank(ImPageRank)algorithm is designed to identify important nodes in a directed hypergraph.The algorithm introduces the Jaccard similarity of directed hypergraphs.By comparing the numbers of common neighbors between nodes with the total number of their neighbors,the Jaccard similarity measure takes into account the similarity between nodes that are not directly connected,and can reflect the potential correlation between nodes.An improved susceptible-infected(SI)model in directed hypergraph is proposed,which considers nonlinear propagation mode and more realistic propagation mechanism.In addition,some important node evaluation methods are transferred from undirected hypergraphs and applied to directed hypergraphs.Finally,the ImPageRank algorithm is used to evaluate the performance of the SI model,network robustness and monotonicity.Simulations of real networks demonstrate the excellent performance of the proposed algorithm and provide a powerful framework for identifying important nodes in directed hypergraphs.

The Internet of Things (IoT) requires wireless communications with low-complexity terminals and high-capacity multiple-access capability. This article focuses on the downlink precoder design for multiuser multiple-input-multiple-output (MIMO) systems with only magnitude channel state information (mCSI). The system applies a complementary block coding {s, 1s) at the transmitter and an energy difference detector at the receivers. With the equivalent channel model of the proposed MIMO system based on noncoherent reception, a zeroforcing precoding scheme is presented depending on the mCSI instead of the full channel information. Further derivation shows that the statistical channel covariance matrix can replace this mCSI to achieve a stable precoding scheme. Through postprocessing signal-to-interference-plus-noise ratio (SINR) analysis, an optimized precoding waveform is derived. Under total power constraints, the obtained SINR processing gain increases non-linearly with the number of transmit antennas, but this gain is not proportional as in coherent MIMO. In addition, a user selection strategy is proposed to mitigate the effect of channel correlation on precoding, and the overall SINRs are proved to be significantly improved. Numerical results verify the spatial-division multiple access (SDMA) of the noncoherent precoding and the effectiveness of the proposed user selection scheme.

The noncoherent single-input-multiple-output (SIMO) system experiences severe cross-product mutual interference with a limited number of receive antennas, severely limiting the number of simultaneous users that can be served in the uplink. In Rician channel, a multiuser energy difference detection (EDD) scheme is introduced for dual-frequency complementary amplitude signals. It eliminates cross-product interference, significantly reducing the number of antennas required by the base station (BS), while extending the constellation space for energy detection (ED). Based on this extended constellation space and channel gain information, two optimized multiuser constellations are obtained by maximizing the minimum joint constellation distance and directly equalizing the receive constellation space. Additionally, the impacts of user mobility are analyzed, and a simple strategy is designed to maintain performance by updating the moving user’s path gain. Simulation results show that the proposed scheme can handle more simultaneous users with the same scale antennas, and it also exhibits strong robustness to user mobility at the same time.

Human digital twin (HDT) is envisioned as a system interconnecting physical twins (PTs) in the real world with virtual twins (VTs) in the digital world, enabling advanced human-centric applications. Unlike optimizing the quality of experience (QoE) of users in single-modal signal transmission for conventional services, users’ QoE in multi-modal signal transmission required by HDT is difficult to guarantee. To tackle this, we study an optimization of QoE in multi-modal transmission, focusing on joint visual and haptic signal feedback transmissions from VT to its PT, for providing immersive interactions in HDT. To evaluate a synthesized performance of visual and haptic experiences, we design a comprehensive QoE model, taking into account video quality, continuous video quality switching rate and average haptic feedback error. Then, to maximize QoE with a guarantee on synchronization between visual and haptic signal transmissions, we dynamically optimize bandwidth allocation, bitrate and rendering mode of the video, and haptic signal’s compression threshold. To this end, we propose a deep reinforcement learning based algorithm, called VisHap. Furthermore, we build an HDT multi-modal interaction platform for collecting an authentic dataset, and by using it, we conduct experiments, showing that VisHap is not only feasible but also outperforms the counterparts.

This paper proposes a new complex dynamical network defined on multi-directed hypergraphs, and designs an adaptive pinning control scheme to regulate the network to achieve complete synchronization. By using the Lyapunov function method and analyzing the defined Laplacian matrices of multi-directed hypergraphs, we obtain the sufficient condition for global asymptotical stability of the synchronization of controlled dynamical network. Based on the obtained stability condition and renumbering theorem, we develop novel algorithms to determine pinned nodes under different directed higher-order structures. By introducing directed weighted hypergraph models, we further find that the directionality of multi-directed hypergraphs can either promote synchronization or suppress synchronization, depending on the weighting method and coupling strength. By utilizing the chaotic Chua’s circuit as nodes’ local dynamics, some numerical examples are presented to illustrate the theoretical results.

To address the high computational performance demands of delay-sensitive tasks in complex terrains, the collaborative computation offloading scheme for reusable tasks in mobile edge computing with the assistance of Unmanned Aerial Vehicle (UAV) is proposed. Firstly, the minimization of the average total energy consumption is formulated by jointly optimizing user offloading, user transmission power, server assignment on UAV, computation frequencies of users and UAV servers, as well as UAV flight trajectory, while meeting the latency constraints. Secondly, a deep reinforcement learning approach is employed to solve the optimization problem, and a Soft Actor-Critic (SAC) based optimization algorithm is introduced. The SAC algorithm utilizes a maximum entropy policy to encourage exploration that enhances the algorithm’s exploration capabilities and accelerates the training convergence speed. Simulation results demonstrate that the proposed SAC algorithm effectively reduces the average total energy consumption of the system while exhibiting good convergence.

Aiming at the inflexiability of network traffic monitoring and tardy response to network emergencies for LS-LEO(large-scale low earth orbit)satellite networks,a method of network traffic monitoring for LS-LEO satellite networks based on INT(in-band network telemetry)is proposed.In order to reduce the overhead of LS-LEO satellite networks traffic monitoring,the proposed method combines the high dynamic topology characteristics of LEO satellite networks,and provides aggregation node selection of network traffic monitoring information within each domain of LEO satellite networks and leader node selection based on microburst traffic,which realizes the rapid monitoring and reporting for abnormal state information of network traffic.In the domain-based aggregation node selection strategy,the aggregation node is obtained by calculating the shortest path and delay.The microburst node is monitored by the telemetry information reporting strategy of the leader node,and the telemetry information is directly reported by the node to the aggregation node.The simulation results show that,compared with the traditional traffic monitoring method,the proposed method can effectively reduce the overhead of telemetry packets and realize flexible low-overhead monitoring of bursts in traffic.Taking the LEO satellite network scale of 500 nodes as an example,the average telemetry overhead for a flow with a path length of 7 hops can be reduced by 27.77%.