A Variational Bayesian Perspective on Massive MIMO Detection
A Variational Bayesian Perspective on Massive MIMO Detection
复制标题
大规模 MIMO 检测的变分贝叶斯视角
DOI：
--
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
A. L. Swindlehurst
中科院分区：
文献类型：
--
作者：
D. Nguyen;Italo Atzeni;A. Tõlli;A. L. Swindlehurst
在大规模多输入多输出（MIMO）系统中，最优数据检测需要极高的计算复杂度。文献中提出了各种检测算法，在复杂性和检测性能之间提供了不同的权衡。在本文中，我们基于变分贝叶斯（VB）推理来设计用于大规模MIMO系统的低复杂度多用户检测算法。我们首先研究了接收端具有完美信道状态信息（CSIR）的大规模MIMO检测问题，并表明具有已知噪声方差的传统VB方法的检测性能较差。为了解决这一限制，我们设计了两种新的VB算法，它们使用算法本身假设的噪声方差和协方差矩阵。我们进一步开发了VB框架，用于不完善的CSIR下的大规模MIMO检测。仿真结果表明，在广泛的信道模型下，与现有方案相比，所提出的VB方法的检测误差明显降低。
Optimal data detection in massive multiple-input multiple-output (MIMO) systems requires prohibitive computational complexity. A variety of detection algorithms have been proposed in the literature, offering different trade-offs between complexity and detection performance. In this paper, we build upon variational Bayes (VB) inference to design low-complexity multiuser detection algorithms for massive MIMO systems. We first examine the massive MIMO detection problem with perfect channel state information at the receiver (CSIR) and show that a conventional VB method with known noise variance yields poor detection performance. To address this limitation, we devise two new VB algorithms that use the noise variance and covariance matrix postulated by the algorithms themselves. We further develop the VB framework for massive MIMO detection with imperfect CSIR. Simulation results show that the proposed VB methods achieve significantly lower detection errors compared with existing schemes for a wide range of channel models.