Unmanned Aerial Vehicles (UAVs) have shown great application potential in the low-altitude economy, intelligent logistics, and emergency communication. In this paper, we first replace traditional ground APs with UAVs as aerial access points in a Cell-Free massive MIMO (CF-mMIMO) system. Subsequently, some UAVs are equipped with Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RISs) to form a STAR-RIS-UAV-CF-mMIMO architecture, achieving full-domain coverage and efficient resource coordination. Different from existing research, in this paper, we simulate user heterogeneity by using different numbers of antennas, and then consider the differentiated Quality of Service (QoS) requirements of heterogeneous users in the CF-mMIMO system. Based on this, we propose a DDPG-based Resource Allocation for STAR-UAV-RIS-CF-mMIMO System (DRA-SUCS) scheme. This scheme aims to optimize the position coordinates of UAVs, the phase shift matrix and reflection amplitude coefficients of STAR-RISs, as well as the transmit and receive beamforming matrices, so as to maximize the information transmission rate in this system. Simulation results show that this scheme significantly improves the total information transmission rate of the STAR-RIS-UAV-CF-mMIMO communication system through effective resource allocation.
This paper presents the design, fabrication, and characterization of a dual-polarized square patch antenna implemented on a low-temperature co-fired ceramic (LTCC) substrate for integration into 6G upper-mid band Antenna-in-Package (AiP) modules targeting mobile devices. To ensure accurate modeling and optimized antenna performance, the customized LTCC substrate was first fabricated and comprehensively characterized. In particular, the dielectric con-stant and loss tangent were extracted using the T-resonator method, enabling precise parameterization for electromagnetic simulations and subsequent anten-na design. Based on the characterized substrate properties, a compact dual-polarized patch antenna was designed to cover the 14.9–15.9 GHz band, which is considered a promising allocation for 6G upper-mid band applications. The antenna employs orthogonal feed structures to achieve dual polarization with excellent isolation. The fabricated prototype demonstrates a port isolation better than 20 dB across the operating band, confirming the effectiveness of the feed and layout configuration. Measured results validate the simulated predictions, with realized gains of 5.8 dBi and 6.1 dBi for the two orthogonal polarizations, respectively. The radiation patterns exhibit good symmetry and stability, sup-porting reliable performance in polarization-diverse communication scenarios. The compact profile, combined with the inherent multilayer integration capabil-ity of LTCC, makes the proposed antenna well-suited for AiP modules in next-generation mobile platforms. Overall, this work establishes an effective design methodology that links material characterization with antenna implementation on LTCC, paving the way for high-performance, dual-polarized antennas in future 6G AiP modules
As augmented reality and multi-user collaborative applications develop, the interaction between users becomes highly dynamic and complex. In the traditional research on data volume allocation, a static user relationship modeling method is commonly used. That is a fixed topology or average interaction strength is predefined and remains constant throughout the optimization process. However, this method has difficulty reflecting the time-varying characteristics of users during the actual movement and interaction process. To address this issue, this paper proposes a data allocation scheme that combines graph attention networks and Stackelberg game theory optimization. This scheme first uses Graph Attention Networks to model the trajectory features of users, dynamically extracting the interaction weights at each moment to calculate the strength of relationships between users. Then, three types of utility functions are established under the game framework. Under the constraint condition of incentive compatibility, the optimal data allocation results at each moment are obtained by optimization. The results show that the scheme can adaptively adjust the data allocation strategy according to the dynamic interactive weight. The system’s total utility is significantly improved, and the incentive compatibility of multiple participants is ensured, which proves its effectiveness and superiority in dynamic augmented reality environment.
Neural implicit representations enhance model expressiveness by learning encodings in high-dimensional feature spaces and have found widespread applications across various domains. In recent years, the integration of neural implicit representations with Simultaneous Localization and Mapping (SLAM) has seen substantial progress. However, such methods often suffer from low localization accuracy, leading to ghosting artifacts and blurring in reconstructions. These issues are particularly pronounced in large-scale scenes, where they exhibit severe catastrophic forgetting, substantial memory consumption, and difficulties in achieving rapid processing. To accomplish high-quality real-time reconstruction, we process large-scale scenes in a hierarchical manner and incorporate an improved neural implicit SLAM algorithm utilizing specific feature points. The selected keyframes promote the convergence of the entire model, and accelerate the operational speed of the SLAM system while maintaining high reconstruction accuracy and tracking quality.
Multi-modal collaborative perception is essential for robotic systems such as autonomous vehicles. However, the interaction of multi-modal data tends to be bandwidth-intensive, making the trade-off between perception accuracy and communication overhead a critical challenge in collaborative perception. To address this issue, we propose Fusion-based semantic Communication (FuseComm), a framework in which both video and LiDAR data are encoded using joint source-channel coding (JSCC). The resulting encoded symbols are transmitted to an edge service for multi-modal fusion and object detection. A global-local fusion mechanism integrates global context via cross-attention and refines local features through a position-guided strategy, ensuring robust fusion even when inputs are corrupted. Extensive experiments demonstrate that FuseComm significantly outperforms LiDAR-based systems under imperfect channel conditions, achieving substantial gains in detection accuracy.
