Jun-Bo Wang

With the rapid development of low earth orbit (LEO) satellites, the fifth generation (5G) satellite-terrestrial integrated network (STIN) has been proposed recently to provide global coverage and universal multi-access. However, the complex communication environments of the 5G STIN, such as dynamic topology and intermittent connectivity, pose challenges in designing handover schemes. This paper investigates the satellite-ground handover problem of the 5G STIN. Initially, we consider a 5G STIN, where multiple LEO satellites and multiple ground base stations are taken as candidate nodes to provide Internet access for a ground user. We establish the satellite channel model and the ground channel model, respectively. To make accurate handover decisions, we consider three attributes, i.e., the received signal-to-noise ratio, the remaining service time, and the remaining load. Then, we analyze the composite attribute value by the entropy weight method and the composite link quality by the adaptive modulation and coding scheme. We formulate a Markov decision process based satellite-ground handover (MDPSGH) problem by maximizing the expected total reward. However, due to the large state space and action space, the MDPSGH problem suffers from the curse of dimensionality. To tackle such a problem, we divide it into a non-terrestrial candidate node (NTCN) selection subproblem and a dimension-reduced MDPSGH subproblem. We propose an entropy weight method-based NTCN selection algorithm and a Bellman optimality equation-based value iteration algorithm to solve these two subproblems. After that, we construct the overall MDPSGH algorithm and analyze its computational complexity as well as convergence. Simulation results show that the proposed MDPSGH algorithm achieves the best performance compared with the existing benchmarks.

Wireless virtual reality (VR) demands ultra-high resolution, ultra-low latency, and real-time computationally intensive viewport rendering, which jointly impose stringent requirements on communication and computing. To enhance the quality of experience for VR users, this paper investigates a mobile edge computing assisted VR streaming system employing extremely large-scale multiple-input multiple-output (XL-MIMO) at the base station (BS). Considering smoothness, clarity, and duration of the VR experience, we formulate a joint optimization problem that minimizes the weighted energy consumption by optimizing the transmit association between VR users and BS subarrays, quantization parameter (QP), rendering offloading decision, beamforming vector, and computing resource allocation. To tackle the high-dimensional mixed-integer non-convex problem efficiently, we propose a hierarchical collaborative optimization based joint rendering offloading and resource allocation (HC-JRORA) algorithm. The upper layer leverages traditional convex optimization methods to determine the computing resource allocation, QP selection, and beamforming design, while the lower layer optimizes transmit association and rendering offloading decision based on a dueling double deep Q network (D3QN) enhanced with imitation learning. Simulation results show that the proposed HC-JRORA algorithm can achieve lower weighted energy consumption and can provide optimized solutions faster than traditional iterative optimization methods.

In near-field multiple-input multiple-output (MIMO) communications, the arbitrary orientation of antenna arrays introduces unique challenges to beam training, as energy distributions become highly dependent on array configurations and exhibit complex spatial patterns. To address these challenges, we present a unified dual-stage two-dimensional beam training framework that optimizes both angular coverage and beamwidth so as to adapt to varying channel characteristics of the two-sided near-field channels. We introduce a two-stage adaptive beam training framework based on a generalized angular-domain channel representation that decomposes array responses into position-dependent and structure-dependent components, revealing how spatial energy distributions evolve with array orientations. Specifically, in Stage I, we extract channel energy distribution patterns through initial beam sweeping. The adaptive beam training scheme in Stage II consists of three strategies: hierarchical refinement for concentrated energy scenarios, angular rotation for transitional energy distributions, and beamwidth expansion for dispersed energy patterns. This unified framework integrates these techniques into a coherent system, adapting to the full spectrum of near-field channel conditions without prior knowledge of array configurations. Extensive simulations demonstrate that our framework significantly outperforms conventional methods in achievable rate and reliability metrics.

The deployment of extremely large antenna arrays (ELAAs) in extremely large-scale multiple-input multiple-output (XL-MIMO) systems introduces significant near-field effects, such as spherical wavefront propagation and spatially non-stationary (SnS) properties. When combined with the dual-wideband effects inherent to wideband systems, these phenomena fundamentally alter the channel’s sparsity patterns in the angular-delay domain, rendering existing estimation methods insufficient. To address these challenges, this paper reconsiders the channel estimation problem for wideband XL-MIMO systems. Leveraging the spatial-chirp property of array responses, we first quantitatively characterize the angular-delay domain sparsity of wideband XL-MIMO channels, revealing both global block sparsity and local common-delay sparsity. To effectively capture this structured sparsity, we then propose a novel column-wise hierarchical prior model that integrates a precision sharing mechanism and a Markov random field (MRF) structure. Building on this prior model, the channel estimation task is formulated as a multiple measurement vector (MMV)-based Bayesian inference problem. Tailored to the complex factor graph induced by this hierarchical prior, we develop a MMV-based hybrid message passing (MMV-HMP) algorithm. This algorithm performs message updates along the edges of the factor graph, and selectively applies either the variational message passing (VMP) or sum-product (SP) rules, depending on the factor-node structure and message tractability. Simulation results validate the effectiveness of the proposed column-wise hierarchical prior model through ablation studies and demonstrate that the MMV-HMP algorithm, while maintaining moderate computational complexity, consistently outperforms existing baselines which fail to capture the structured sparsity of wideband XL-MIMO channels.

This paper investigates the spatial bandwidth of line-of-sight (LoS) channels in extra-large MIMO (XL-MIMO) systems. For linear large-scale antenna arrays (LSAAs) with transceivers randomly positioned in 3D space, a simple but accurate closed-form expression is derived to characterize the local spatial bandwidth. Based on this analysis, we examine the properties of local spatial bandwidth and further derive expressions for the effective spatial bandwidth and the achievable degrees of freedom (i.e., the K number) for LSAAs. We also conduct case studies for both coplanar and non-coplanar transmitting and receiving arrays, providing more concise and intuitive expressions for local spatial bandwidth and achievable spatial degrees of freedom. Finally, the impact of array geometry on LoS XL-MIMO channel capacity is explored. When the transmitting and receiving arrays are coplanar and perpendicular to the line connecting their centers, the effective degree of freedom of the LoS channel is found to be approximately maximized. This orientation also maximizes the channel capacity in near-field high-SNR scenarios.

In this paper, we investigate the channel estimation problem for extremely large-scale multiple-input multiple-output (XL-MIMO) systems with a hybrid analog-digital architecture, implemented within a decentralized baseband processing (DBP) framework with a star topology. Existing centralized and fully decentralized channel estimation methods face limitations due to excessive computational complexity or degraded performance. To overcome these challenges, we propose a novel two-stage channel estimation scheme that integrates local sparse reconstruction with global fusion and refinement. Specifically, in the first stage, by exploiting the sparsity of channels in the angular-delay domain, the local reconstruction task is formulated as a sparse signal recovery problem. To solve it, we develop a graph neural networks-enhanced sparse Bayesian learning (SBL-GNNs) algorithm, which effectively captures dependencies among channel coefficients, significantly improving estimation accuracy. In the second stage, the local estimates from the local processing units (LPUs) are aligned into a global angular domain for fusion at the central processing unit (CPU). Based on the aggregated observations, the channel refinement is modeled as a Bayesian denoising problem. To efficiently solve it, we devise a variational message passing algorithm that incorporates a Markov chain-based hierarchical sparse prior, effectively leveraging both the sparsity and the correlations of the channels in the global angular-delay domain. Simulation results show the effectiveness and superiority of the proposed SBL-GNNs algorithm over existing methods, demonstrating improved estimation performance and reduced computational complexity.