With the rise of low-altitude economy applications, 6G communication systems place stricter demands on optical fiber amplifiers, requiring wider bandwidth, higher gain, and better spectral uniformity. Backbone networks for low-altitude integrated sensing and communication systems, in particular, call for high-performance amplification to support robust data transmission and reliable sensing. However, traditional multi-pump Raman fiber amplifiers (RFAs) are no longer adequate for meeting the performance demands of distributed fiber optic sensing networks in such scenarios. To address this problem, this paper proposes a machine learning-enhanced multi-pump RFA for high-performance optical backbone in low-altitude sensing and communication. The back propagation neural network (BPNN) is employed to accurately model the nonlinear relationship between pump parameters and amplification performance, facilitating adaptive and fine-grained control over signal gain, which is critical for maintaining stable and efficient data transmission across dynamic and heterogeneous communication scenarios. Moreover, the artificial bee colony (ABC) algorithm is integrated to perform global optimization of pump wavelengths and power configurations, thereby improving overall system bandwidth, gain characteristics, and operational robustness under diverse and unpredictable network conditions. The experimental results demonstrate that the proposed method achieves superior prediction accuracy, enhanced stability, and greater adaptability compared to conventional algorithms.

With the rapid growth of the low-altitude economy, emerging applications such as urban air mobility and smart logistics demand reliable and low-latency beamforming for unmanned aerial vehicle-to-vehicle (UAV-to-UAV, U2U) communications in millimeter-wave (mmWave) bands under highly dynamic and non-line-of-sight (NLOS) conditions. Traditional beam alignment methods relying on exhaustive search or channel feedback incur heavy training overhead and degraded accuracy in rapidly varying environments. To address these challenges, we propose multi-modal perception-assisted fusion network (MPFusionNet), a multi-modal perception-enhanced Transformer framework for predictive beamforming. Our approach leverages heterogeneous onboard sensing data including global positioning system (GPS), red-green-blue (RGB) cameras, LiDAR, and radar altimeters, incorporates a dynamic time warping (DTW)-based alignment mechanism, and embeds geometry-aware priors within a perceiver input-output (PerceiverIO)-based fusion architecture to achieve robust spatiotemporal representation. Experiments on a simulated U2U dataset show that MPFusionNet attains a top-3 beam prediction accuracy of 97.59%, substantially surpassing conventional models. These results demonstrate the effectiveness of multi-modal learning in improving robustness and generalization of predictive beamforming for future autonomous aerial communication systems.

Extensive channel measurement campaigns have shown that multipath components (MPCs) generally exhibit clustered distributions, making cluster-based models a cornerstone of wireless channel modeling. Developing time-varying cluster-based channel models requires not only clustering MPCs in the delay, angle, and power domains, but also tracking their temporal evolution. However, existing multipath clustering algorithms typically rely on fixed hyperparameters and therefore are often poorly suited to dynamically evolving clusters, while tracking algorithms based solely on distance metrics are prone to ambiguous associations among spatially proximate clusters. Moreover, treating clustering and tracking as separate processes often prevents temporal evolution information from being fully exploited during clustering. This paper proposes a structure-aware unified framework that integrates Adaptive Neighborhood Robust Mean Shift (AN-RMS) clustering with BoxKF-Cluster tracking. Within the proposed framework, AN-RMS, built upon kernel density estimation, adaptively determines the effective number of nearest neighborsKby detecting abrupt changes in the second-order gradient of the neighborhood-distance sequence, thereby identifying structural cluster boundaries and providing locally adaptive guidance for density-mode iteration. BoxKF-Cluster introduces oriented bounding boxes derived from root-mean-square statistics to characterize the evolving morphology of clusters, and combines Kalman filtering with a successive-interference-cancellation (SIC)-like strategy, in which the influence of existing clusters is first removed to facilitate the detection of newly emerging ones. As a result, clustering and tracking are jointly accomplished within a unified framework. The proposed method is validated in complex scattering scenarios using both ray-tracing simulations and real vehicle-to-vehicle millimeter-wave measurement data, demonstrating its effectiveness for time-varying channel modeling.

Integrated Sensing and Communication (ISAC) systems represent a transformative paradigm for next-generation wireless networks by enabling dual-functional efficiency through simultaneous information transmission and environmental sensing. This paper investigates the critical challenge of robust beamforming design for monostatic ISAC systems operating in the near-field (NF) regime, where conventional far-field channel assumptions become fundamentally invalid. We develop a novel robust beamforming framework that optimizes minimum mean squared error (MMSE) estimation for sensing performance while guaranteeing stringent communication quality-of-service requirements. A distinctive feature of our approach lies in the proposed spherical wavefront-based channel model that incorporates both distance and angular response, providing superior accuracy compared to conventional planar wavefront approximations in NF scenarios. To resolve the inherent non-convex optimization problem with coupled sensing-communication constraints, we devise an efficient semidefinite relaxation (SDR)-based algorithm with guaranteed convergence properties. Comprehensive simulations demonstrate significant improvements in both sensing and communication performance, even under imperfect channel state information.

Integrated sensing and communication (ISAC) sys-tems have attracted considerable attention in recent years. This paper proposes an ISAC system based on orthogonal time frequency space (OTFS) modulation, designed to enhance performance in high-mobility environments. We introduce a novel passive sensing method that enables high-resolution target parameter estimation through fine-grained grid search, two-stage parameter refinement. Additionally, we develop a channel estimation method that leverages sensing parameters, utilizing delay-Doppler domain information in OTFS systems to enhance accuracy in high-mobility scenarios. Simulation results across various Signal-to-Noise ratio (SNR) conditions demonstrate the effectiveness of the proposed methods, showing a significant reduction in the root mean square error (RMSE) of distance and velocity estimations as SNR increases. These findings highlight the accuracy of the proposed algorithms in high-noise environments.

ABSTRACT With the evolution of wireless communication technologies towards the sixth generation (6G) mobile communication system, the space‐air‐ground‐sea integrated network architecture has emerged as a critical development direction for achieving global seamless coverage. Focusing on the unmanned aerial vehicle (UAV)‐to‐ship maritime communication scenario within this network framework, a three‐dimensional (3D) geometry‐based stochastic model is proposed. The model adopts a combined structure of elliptical and cylindrical components to comprehensively characterize multipath propagation mechanisms, including line‐of‐sight, sea surface reflection, as well as single‐bounced and double‐bounced components. By introducing the wave equation of sea surface to establish the 3D motion trajectory model of the ship and integrating it with the 3D rotational motion model of the UAV, the time‐varying propagation distance‐induced channel non‐stationarity is accurately captured. Based on this model, key statistical characteristics such as the space‐time‐frequency correlation function (STF‐CF) and Doppler power spectral density are derived. Furthermore, the impacts of sea surface wind speed, UAV rotation, ship oscillation, and ship size on channel statistical properties and space‐time non‐stationarity are thoroughly analysed. These numerical results provide theoretical foundations for the design and performance optimization of UAV‐assisted communication systems in complex maritime environments.

Traffic flow prediction (TFP) plays a crucial role in optimizing road resource allocation and alleviating traffic congestion. However, existing TFP methods have limitations in capturing the complex spatiotemporal dependencies from traffic data, resulting in low prediction accuracy. To solve this problem, we propose a hybrid long short-term memory (LSTM) and graph attention network (GAT) model based on multi-scale temporal features (MSTF-LG) to predict traffic flow. Firstly, we employ trigonometric functions (TF) to process timestamp information to extract its periodicity and continuity features. These features are then integrated with traffic data to construct a comprehensive input representation. We further extract recent traffic data as well as daily, weekly, and monthly periodic data from this representation. Secondly, we adopt the LSTM encoder to process recent traffic data to extract recent trend features, and apply LSTM encoders to handle daily, weekly and monthly periodic data to extract periodic features. Furthermore, we employ the GAT to process these LSTM-encoded multi-scale temporal features of traffic data to capture dynamic spatial characteristics. The normalized GAT outputs are fed into the LSTM decoder to effectively capture the dynamic temporal changes of traffic data, and then a linear layer transforms the output of the LSTM decoder into TFP results. Finally, experimental results demonstrate that the proposed method outperforms existing TFP methods in terms of prediction accuracy, robustness and computational efficiency.

Unmanned aerial vehicles (UAVs), known for their high flexibility and maneuverability, are regarded as the aerial platforms of future integrated sensing and communication (ISAC) networks. The communication and sensing functions of ISAC share the same spectrum and signal waveform, which often results in communication information being embedded within the sensing waveforms, thereby increasing the risk of information leakage. To enhance the security of UAV-assisted ISAC, we propose a beamforming strategy based on the mutual cooperation between communication and sensing. Specifically, by utilizing the sensing function to process echo signals, we estimate the positions of potential eavesdroppers and obstacles, which supports subsequent obstacle avoidance trajectory planning and physical layer security design. To ensure the transmission secrecy, we introduce artificial noise into the system. By designing the UAV transmit beamforming and the covariance matrix of the artificial noise, we formulate an optimization problem that aims to minimize the signal-to-noise ratio (SNR) received by the eavesdropper. To address this non-convex optimization problem, we propose an optimization algorithm that combines Dinkelbach's transform and semidefinite relaxation (SDR). Simulation results demonstrate that the SNR of eavesdropper remains at a low level throughout the flight of UAV, validating the effectiveness of the proposed scheme.

The cloud-edge-end collaboration framework is emerging as a promising means to handle diverse tasks and improve Quality of Service. Existing research rarely considers the affinity between diverse tasks and heterogeneous resources, preventing the further improvement of system performance. This paper proposes a dynamic task scheduling approach based on the intelligent affinity strategy for cloud-edge-end collaboration. The affinity strategy depicts the preference of tasks for computing nodes with different labels, including resource types and node zones, and each task can set multiple affinity rules to match target nodes. Additionally, the proposed approach adopts deep reinforcement learning theory to generate the affinity rules, which means utilizing an intelligent algorithm to guide the rule generation. Extensive experiments show that the proposed approach can reduce the average cost by at least 20% compared with the baseline.

The internet of vehicles (IoV) is becoming the key scenario in the current and future intelligent applications, pro-pelled by the advancements of communication technology and integration of artificial intelligence and the autonomous systems. Realizing the reliable data transmission in channel variability and estimation inaccuracies situation in such high-mobility vehicular networks is urgent. The orthogonal time frequency space (OTFS) can be utilized to achieve this destination. Therefore, we introduce an OTFS signal detection method using a residual channel attention network (RCAN), applied in a data-driven paradigm, recovers input signals from received information in an end-to-end manner, reducing the need for explicit channel estimation. The RCAN we designed substantially enhances the network's learning capabilities and optimizes its overall performance. Differentiate from the classic deep learning network-based OTFS signal detection algorithms, the simulation results reveal that the proposed RCAN-based OTFS signal detection scheme has a lower bit error rate (BER) under the same SNR situation, and also has lower space complexity. Furthermore, our method exhibits robustness when the transmission channel changes.

Vehicle position prediction (VPP) is of great significance for navigation planning and traffic safety of intelligent vehicles. In general, particle filtering (PF) uses global navigation satellite system (GNSS) to implement VPP. However, it does not consider geographic layer information (GLI) and its particle weight is not combined with the real-world geographic position information, which leads to insufficient prediction preparation. To resolve this problem, we propose a novel PF-based VPP method by using three-dimensional convolutional neural network and long short-term memory (3D CNN-LSTM) network model. First, for data preprocessing, we extract kinematic information features from GNSS, and evenly divide the area around each GNSS point into multiple grids and calculate the probability of grids center belonging to each GLI type. In addition, in order to better reflect the relationship between two consecutive positions due to the factors such as the conversion angle, we construct tilted cells to represent possible positions of each vehicle at any time. Second, a novel 3D CNN-LSTM model is designed to calculate the vehicle occurrence probability (VOP) in each tilted cell by processing the GLI and GNSS data, which can optimize the PF weight of each particle, and then improve PF to make more precise position prediction. Finally, the experimental results demonstrate that the proposed VPP method can improve the cell prediction accuracy, and then significantly improve the position prediction precision.

Edge computing has become a promising technology to mitigate the latency of various cloud services. Task scheduling in edge computing is challenging due to the heterogeneous devices and multiple tasks. This paper proposes an affinity-based scheduling algorithm to solve the multi-task scheduling problem with heterogeneous computing resources under edge computing. The algorithm uses a centralized scheduling strategy that takes into account the overall matching between intelligent computing tasks and heterogeneous resources from two aspects: resource affinity and system load balancing. It can adjust the weights among these aspects to meet different application requirements. The results show that the proposed algorithm can reduce the average response time by at least 10% and has a significant advantage regarding the system running time compared to the other comparative methods.

Non-linear Multiple-Input Multiple-Output (MIMO) has attracted considerable attention because of its high power-efficiency characteristic, particularly in the fifth generation (5G) and beyond. This paper focuses on the non-linear MIMO baseband algorithms in power-efficiency networks. In previous works, Generalized Approximate Message Passing (GAMP) and importance sampling technique were used to solve the non-linear distortion in Halved Phase-Only (HPO-) MIMO system. However, its convergence rate becomes unstable, and it’s converge is not guaranteed in some cases. In this paper, to improve the efficiency of convergence rate, we propose Gaussian Mixture Model (GMM) based ExpectationMaximization (EM) signal processing algorithm in HPO MIMO system. We first transforme channel estimation and multiuser detection problems into generalized linear mixed problems under π-phase observations. Then, the GMM algorithm is used to estimate the distribution of π-phase observation. Meanwhile, the EM algorithm is used to estimate the recovered signal. Simulation results show that the proposed method achieves high convergence and has better performance than the reference GAMP algorithm.

Massive multiple-input multiple-output (MIMO) wireless communication technology with the characteristics of hyperconnectivity is an ideal channel to connect the industrial Internet of Things (IIoT) and the cyber-physical system. It provides stable and reliable connectivity from the data center to distributed user terminals and the IIoT. However, traditional massive MIMO suffers from high power consumption and fabrication cost. The design of energy-efficient massive MIMO technology is essential for larger scale industrial deployments. In this article, we design three types of nonlinear RF chain structures, which not only reduce the power consumption of massive MIMO systems but also save fabrication costs. Information theoretic analysis demonstrates the power efficiency performance of our nonlinear system design. Our nonlinear MIMO system designs can increase the power efficiency by up to 2.3 times compared with the traditional MIMO system. We have demonstrated that our systems can achieve the same uplink rate as traditional MIMO by increasing the number of receiving antennas but with less overall power consumption. We also proposed an algorithm to overcome the problem of low computational efficiency due to high-dimensional integration when calculating the uplink achievable rate of nonlinear MIMO. Moreover, we reveal that when the skew-normal distribution is used as signaling, the nonlinear MIMO systems can achieve better performance than the Gaussian distribution.

Cooperative localization (CL) based on the angle-of-arrival (AOA) measurements is a promising positioning technique for the wireless sensor network (WSN). This is because CL reaches high localization precision and robustness by exploiting the relative ranging measurements among the agents. In addition, the AOA interpretation involves no propagation parameters, and its acquisition does not require strict time-synchronization among the WSN nodes. In this paper, we firstly categorize the AOA-CL problem as a generalized linear-mixing problem under the phase-only measurements, and then resolve it by our developed phase-only generalized approximate message passing (POG-AMP) algorithm. The POG-AMP localizer is warm-started by developing an incremental localizer assisting by anchor-connectivity and region-boundary constraints. It calls for sparse matrix-vector multiplications (MVMs) as its most complex operations, and can be implemented in a distributed manner. Therefore, it has low computational complexity, and is suitable for WSN hardware implementation. Simulation results validate the state-of-the-art performance and cooperation gains of the POG-AMP localizer.

Cooperative localizer is a potential positioning technique for vehicular ad-hoc networks (VANETs). However, it would suffer from the non-line-of-sight (NLOS) problems widely existing in VANETs. This letter proposes a geographical information enhanced cooperative localizer (GIE-CL) for VANET with time-of-arrival (TOA) measurements. It iterates between NLOS identification and extended generalized approximate message passing (EGAMP) motivated cooperative positioning. A region sampling method is developed to identify NLOS measurements based on geographical information and current vehicle position estimations. Subsequently, the detected NLOS measurements are removed and the EGAMP localizer is activated to re-estimate the vehicle positions. The above-mentioned iteration will be terminated until convergence is reached. Initial positions are provided by Global Navigation Satellite System (GNSS). Simulation results show that GIE-CL can handle the NLOS problem, and approach to its performance upper bound provided by the case with known NLOS/LOS link-type information. Compared to EGAMP localizer, the positioning accuracy of GIE-CL is improved by eight times when the allowed localization error is less than 5 m.