| 摘要: |
| 作为6G的核心使能技术之一，超大规模多输入多输出技术将显著扩展近场区域的范围。近场信道的球面波前特性导致基于压缩感知的信道估计算法性能显著下降。为了解决这一问题，提出了一种联合码本设计方案。联合码本由两部分构建：发送端码本通过对具有中心频率偏移的近场下行信道自相关矩阵进行特征值分解构建，揭示了信道自相关矩阵的特征向量与调制离散椭球谐波序列的紧密关联，提出了一种直接计算信道自相关矩阵的快速方法以实现高效的特征值分解；接收端码本通过加权求和离散傅里叶变换和离散余弦变换码本构建，提出了一种层级优化方法以确定最佳权重因子。仿真结果表明，所提出的联合码本在降低存储复杂度的同时具有更好的信道估计性能和鲁棒性，信道估计的归一化均方误差性能指标相比传统码本平均降低约10 dB。 |
| 关键词: 近场通信 超大规模多输入多输出 信道估计 码本设计 |
| DOI：10.20079/j.issn.1001-893x.250113008 |
|
| 基金项目:重庆市自然科学基金创新发展联合基金（中国星网）（CSTB2023NSCQ-LZX0114） |
|
| A Joint Codebook Design for Downlink Near-field XL-MIMOChannel Estimation |
| GUO Xiaojin,WANG Shengbao,SONG Yunpeng |
| (School of Communications and Information Engineering，Chongqing University of Posts and Telecommunications，Chongqing 400065，China) |
| Abstract: |
| As one of the core enabling technologies for 6G，extremely large-scale multiple-input multiple-output(XL-MIMO) technology will significantly expand the range of near-field regions.The spherical wave properties of the near-field channel lead to a significant performance degradation in channel estimation algorithms based on compressed sensing(CS).To address this issue，a joint codebook design scheme is proposed.The joint codebook is constructed in two parts.The transmit-side codebook is designed through eigenvalue decomposition(EVD) of the near-field downlink channel autocorrelation matrix with a central frequency offset.The tight connection between the eigenvectors of the channel autocorrelation matrix and the modulated discrete prolate spheroidal sequences(MDPSS) is revealed.A fast method for directly calculating the channel autocorrelation matrix is proposed to enable efficient eigenvalue decomposition.The receive-side codebook is constructed through a weighted summation of discrete Fourier transform(DFT) and discrete cosine transform(DCT) codebooks.A hierarchical optimization method is proposed to determine the optimal weighting factors.Simulation results show that the proposed joint codebook reduces storage complexity while achieving better channel estimation performance and robustness.The normalized mean square error performance of channel estimation is，on average，reduced by approximately 10 dB compared with that of traditional codebooks. |
| Key words: near-field communications XL-MIMO channel estimation codebook design |
《电讯技术》编辑部 