一种基于GMD分解的混合预编码算法

GMD-Based Hybrid Precoding algorithm

提出了毫米波大规模MIMO系统中一种利用几何均值分解(GMD)设计模拟和数字预编码矩阵的算法。首先根据初始模拟预编码矩阵列向量相互正交,利用GMD分解计算出模拟预编码矩阵后,取其元素相位构成的矩阵为最终的模拟预编码矩阵,然后设计数字预编码矩阵。仿真结果表明,频谱效率和误码率较正交匹配追踪(OMP)算法分别有大约3dB和2.5dB的增益。

An algorithm utilizing geometric mean decomposition（GMD） is proposed to design analog and digital precoding matrices in Millimeter Wave Massive MIMO systems. According to the columns of the initial analog precoding matrix are orthogonal to each other,the analog precoding matrix is calculated based on GMD, the matrix composed of the phases of the elements in analog precoding matrix is adopted as the ultimate analog precoding matrix. Then the digital precoding matrix is calculated. Simulation results show that the spectral efficiency and the BER performance are improved by 3 dB and 2.5 dB compared with OMP algorithm, respectively.