AI 中文总结
该研究提出通过旋转均匀线性阵列进行高效三维空间功率谱合成的方法，受CT启发合成三维谱，相比其他方法降低采样和移动开销，还给出方向设计准则并优化方向集，能以少量样本重建全空间三维谱。
AI 中文摘要
本文提出一种通过在三维空间中围绕其中心旋转均匀线性阵列（ULA）来进行高效三维空间功率谱合成的方法。受经典计算机断层扫描（CT）启发，ULA在每个旋转角度执行模拟接收合并以产生部分相干和。通过在多次旋转中收集这些和，可通过单个射频（RF）链在线合成完整的三维谱，无需明确获取每个天线信号。根据整体相干和是否可获取，合成可通过对部分谱图像的最小操作或在累积所有相干和后的联合合成来实现。与基于固定均匀平面阵列（UPA）的合并和用于三维立方虚拟阵列的密集单移动天线（MA）采样相比，该方案在保持均匀高角度分辨率的同时，以大幅降低的采样和移动开销实现全空间三维覆盖。其采样几何和顺序方向设计还支持流水线模拟波束形成，降低实际硬件成本。为在无先验环境信息的情况下以原理性方式设计旋转方向，旨在最大化多径分量（MPC）对之间预期最坏情况投影间隔。次要标准是最小化方向轴之间最坏情况投影相关性以减少方向冗余。因此，使用多起始平滑极小极大算法在各向同性矩阵和单位范数约束下针对不同数量的方向优化方向集。数值结果表明，优化后的方向均匀跨越三维空间，并仅使用一小部分空间样本重建接近密集三维立方参考的全空间三维谱。
英文摘要
This paper proposes an efficient three-dimensional (3D) spatial power spectrum synthesis method by rotating a uniform linear array (ULA) about its center in 3D space. Inspired by classical computed tomography (CT), the ULA performs analog receive combining at each rotation angle to produce a partial coherent sum. By collecting such sums over multiple rotations, the full 3D spectrum can be synthesized online via a single radio-frequency (RF) chain, without explicitly acquiring per-antenna signals. Depending on whether the overall coherent sum is accessible, the synthesis is obtained through either a minimum operation over partial spectrum images or joint synthesis after accumulating all coherent sums. Compared with fixed uniform planar array (UPA)-based combining and dense single movable-antenna (MA) sampling for 3D cubic virtual arrays, the proposed scheme achieves full-space 3D coverage with substantially reduced sampling and movement overhead while maintaining uniformly high angular resolution. Its sampling geometry and sequential orientation design also support pipelined analog beamforming, reducing practical hardware cost. To design rotation orientations in a principled manner without prior environmental information, we aim to maximize the expected worst-case projected separation between multi-path component (MPC) pairs. A secondary criterion then minimizes the worst-case projective correlation among orientation axes to reduce orientation redundancy. Accordingly, we optimize orientation sets for different numbers of orientations under isotropic-matrix and unit-norm constraints, using a multistart smooth minimax algorithm. Numerical results show that the optimized orientations uniformly span 3D space and reconstruct full-space 3D spectra close to the dense 3D cubic reference using only a fraction of spatial samples.
Comments13 pages, submitted to journal for publication
