面向XL-MIMO可视区域识别的非均匀空间采样

摘要：超大规模多输入多输出（XL-MIMO）是面向未来6G超级无线宽带和超大规模连接的关键技术，其空间非平稳特性导致局部天线阵列区域可能仅会被部分用户“看到”，称为用户可视区域（VR）。利用VR可实现XL-MIMO低复杂度传输设计，如何识别用户VR是其必要前提。由于用户VR与其空间位置存在天然联系，可通过选择少量用户估计并反馈其所在位置处的VR，然后结合用户位置外推出其余用户VR。该过程可解释为“VR地图”的空间采样与重建，外推效果与采样点位置选择关系密切。为提高采样效率，基于探测与细化相结合的设计理念，提出了一种有限样本下的非均匀空间采样方案，并分析探测细化调控因子的设计方法。仿真结果表明，所提方案相较于传统随机采样具有更高的效率，可显著提升小样本下的VR识别准确性。

关键词：超大规模多输入多输出；空间采样；可视区域识别；探测细化

Abstract: Extra-large massive multiple-input multiple-output (XL-MIMO) is a key technology for future 6G ultra wireless broadband and ultra large-scale connectivity. Its spatial non-stationary characteristic may result in the local antenna array region being only "visible" to some users, known as the user’s visibility region (VR). Utilizing VR can achieve low complexity transmission design for XL-MIMO, while recognizing user’s VR is a necessary prerequisite. Due to the natural connection between user’s VR and their spatial location, a small number of users can be selected to estimate and provide feedback on the VR at their location, and then combined with the user's location to extrapolate the VR information of other users. This process can be explained as spatial sampling and reconstruction of “VR maps”, and the extrapolation effect is closely related to the selection of sampling point positions. In order to improve sampling efficiency, a non-uniform spatial sampling scheme under limited samples is proposed based on the design concept of combining exploration and refinement, and the design method of exploration and refinement control factor in general scenarios is analyzed. The simulation results show that the proposed scheme has higher efficiency compared to traditional random sampling and can significantly improve the accuracy of VR recognition in small samples.

Keywords: XL-MIMO; spatial sampling; visibility region recognition; exploration and refinement