毫米波massive MIMO系统中混合连接的混合预编码设计

Design of hybridly-connected hybrid precoding in millimeter-wave massive MIMO system

为了提高混合连接的混合预编码的频谱效率,首先利用连续干扰消除(SIC)的原理得到理想条件下的最优混合预编码矩阵,然后利用梯度下降理论将最优混合预编码矩阵分解为数字预编码矩阵和模拟预编码矩阵,最后考虑模拟预编码矩阵的恒模约束条件,并以最大化频谱效率为目标利用交替最小化方法优化模拟和数字预编码矩阵。所提出的混合预编码设计算法基于混合连接结构,因而能量效率远优于部分连接的和全连接的混合预编码。同时,该算法不会增加混合连接的混合预编码的硬件复杂度且只少量增加计算量。仿真结果表明,该算法能提升混合连接的混合预编码的频谱效率,特别是当射频(RF)链路数大于数据流数时,频谱效率的提升更加显著。由于分块不需要满足正交性,该算法比现有混合连接的混合预编码更适合实际应用。

In order to improve the spectral efficiency of hybridly-connected hybrid precoding,the optimal hybrid precoding matrix under the ideal conditions was firstly obtained by using the principle of successive interference cancellation(SIC).Secondly,the optimal hybrid precoding matrix was decomposed into the digital precoding matrix and the analog precoding matrix by using the gradient descent theory.Finally,considering the constant modulus constraint condition of the analog precoding matrix,the digital and analog precoding matrices were optimized by using the alternating minimization method aim to maximize the spectral efficiency.Due to the hybridly-connected structure,the proposed hybrid precoding design algorithm is significantly superior to the partially-connected and fully-connected hybrid precoding in terms of the system energy efficiency.Meanwhile,the algorithm does not increase any hardware complexity and only increases a small amount of computation of the hybridly-connected hybrid precoding.Computer simulation results exhibit that the proposed algorithm can improve the system spectral efficiency of the hybridly-connected hybrid precoding,and the upgrade of spectral efficiency is more significant especially in the conditions that the number of radio frequency(RF)links is greater than the number of data streams.Since the sub-blocks are not necessary to satisfy orthogonality conditions,the proposed algorithm is more suitable for practical application than the existing hybridly-connected hybrid precoding.