The deployment of extremely large antenna arrays (ELAAs) and operation at higher frequency bands in wideband extremely large-scale multiple-input-multiple-output (XL-MIMO) systems introduce significant near-field effects, such as spherical wavefront propagation and spatially non-stationary (SnS) properties. Combined with dual-wideband impacts, these effects fundamentally reshape the sparsity patterns of wideband XL-MIMO channels in the angular-delay domain, making existing sparsity-based channel estimation methods inadequate. To address these challenges, this paper revisits the channel estimation problem for wideband XL-MIMO systems, considering dual-wideband effects, spherical wavefront, and SnS properties. By leveraging the spatial-chirp property of near-field array responses, we quantitatively characterize the sparsity patterns of wideband XL-MIMO channels in the angular-delay domain, revealing global block sparsity and local common-delay sparsity. Building on this structured sparsity, we formulate the wideband XL-MIMO channel estimation problem as a multiple measurement vector (MMV)-based Bayesian inference task and propose a novel column-wise hierarchical prior model to effectively capture the sparsity characteristics. To enable efficient channel reconstruction, we develop an MMV-based variational message passing (MMV-VMP) algorithm, tailored to the complex factor graph induced by the hierarchical prior. Simulation results validate the proposed algorithm, demonstrating its convergence and superior performance compared to existing methods, thus establishing its effectiveness in addressing the challenges of wideband XL-MIMO channel estimation under complex near-field conditions.

翻译：在宽带超大规模多输入多输出（XL-MIMO）系统中，超大规模天线阵列（ELAAs）的部署与高频段运行引入了显著的近场效应，如球面波前传播与空间非平稳（SnS）特性。这些效应与双宽带影响相结合，从根本上重塑了宽带XL-MIMO信道在角度-时延域的稀疏性模式，使得现有的基于稀疏性的信道估计方法不再适用。为应对这些挑战，本文重新审视了宽带XL-MIMO系统的信道估计问题，综合考虑了双宽带效应、球面波前与SnS特性。通过利用近场阵列响应的空间啁啾特性，我们定量刻画了宽带XL-MIMO信道在角度-时延域的稀疏性模式，揭示了全局块稀疏性与局部公共时延稀疏性。基于这种结构化稀疏性，我们将宽带XL-MIMO信道估计问题表述为基于多测量向量（MMV）的贝叶斯推断任务，并提出了一种新颖的列式分层先验模型，以有效捕捉稀疏性特征。为实现高效的信道重构，我们开发了一种基于MMV的变分消息传递（MMV-VMP）算法，该算法专门针对由分层先验导出的复杂因子图进行定制。仿真结果验证了所提算法的有效性，展示了其收敛性及相较于现有方法的优越性能，从而确立了其在复杂近场条件下应对宽带XL-MIMO信道估计挑战的有效性。