Fluid antenna systems (FAS) provide continuous spatial degrees of freedom for next-generation wireless networks. However, deploying FAS on high-speed mobile embodied intelligent agents introduces Doppler shifts that shorten the channel coherence time, causing conventional full-port scanning to incur excessive pilot overhead. To address this issue, this paper proposes a prediction-aided adaptive communication architecture that constructs a dynamic local sounding window centered on upper-layer prediction results. By exploiting the spatial channel correlation among FAS ports, the proposed approach balances pilot overhead and spatial diversity gain under prediction deviations. We establish a quantitative model for the long-term expected throughput by integrating a nonlinear bounded hit probability function and a gain bridging model based on the zeroth-order Bessel function of the first kind. We then formulate a discrete optimization problem under practical constraints to maximize the expected transmission capacity, and develop a two-stage low-complexity algorithm that decouples the online computational complexity from the number of antenna ports. Simulation results show that the proposed mechanism adaptively adjusts the sounding window size according to moving velocity and prediction accuracy, reducing pilot overhead in high-mobility scenarios and achieving higher throughput than fixed-window and global scanning schemes.

Although federated learning (FL) has become very popular recently, it is vulnerable to gradient leakage attacks. Recent studies have shown that attackers can reconstruct clients’ private data from shared models or gradients. Many existing works focus on adding privacy protection mechanisms to prevent user privacy leakages, such as differential privacy (DP) and homomorphic encryption. These defenses may cause an increase in computation and communication costs or degrade the performance of FL. Besides, they do not consider the impact of wireless network resources on the FL training process. Herein, we propose weight compression, a defense method to prevent gradient leakage attacks for FL over wireless networks. The gradient compression matrix is determined by the user’s location and channel conditions. We also add Gaussian noise to the compressed gradients to strengthen the defense. This joint learning of wireless resource allocation and weight compression matrix is formulated as an optimization problem with the objective of minimizing the FL loss function. To find the solution, we first analyze the convergence rate of FL and quantify the effect of the weight matrix on FL convergence. Then, we seek the optimal resource block (RB) allocation by exhaustive search or ant colony optimization (ACO) and then use the CVX toolbox to obtain the optimal weight matrix to minimize the optimization function. The simulation results show that the optimized RB can accelerate the convergence of FL.

Reconfigurable intelligent surface (RIS) as a promising technology has been proposed to change weak communication environments. However, most of the current resource allocation (RA) schemes have focused on RIS-assisted homogeneous networks, and there is still no open works about RA schemes of RIS-assisted heterogeneous networks (HetNets). In this paper, we design an RA scheme for a RIS-assisted HetNet with non-orthogonal multiple access to improve spectrum efficiency and transmission rates. In particular, we jointly optimize the transmit power of the small-cell base station and the phase-shift matrix of the RIS to maximize the sum rates of all small-cell users, subject to the unit modulus constraint, the minimum signal-to-interference-plus-noise ratio constraint, and the cross-tier interference constraint for protecting communication quality of microcell users. An efficient suboptimal RA scheme is proposed based on the alternating iteration approach, and successive convex approximation and logarithmic transformation approach. Simulation results verify the effectiveness of the proposed scheme in terms of data rates.