单用户毫米波大规模多输入多输出系统的几何均值混合预编码

针对奇异值分解在不同的信道及信噪比中进行比特分配造成的误码问题,提出了一种基于几何均值分解的混合预编码算法。在同时考虑预编码器和合成器的情况下,首先在基站端用奇异值分解设计数字编码器,利用基追踪原理设计模拟预编码器。然后在接收端采用启发式算法来设计数字合成器和模拟合成器。仿真结果表明,提出的混合预编码方案比传统基于奇异值的混合预编码具有更低的误码率,且频谱效率更接近理想值。

Single User Geometric Mean Decomposition Hybrid Precoding for massive MIMO Millimeter-Wave System

Aiming at the bit error rate caused by bit allocation in singular value decomposition in different channels and SNR, a hybrid precoding algorithm based on geometric mean decomposition is proposed. In the case of considering both the precoder and the synthesizer, the digital encoder is first designed with singular value decomposition at the base station side, and the analog precoder is designed using the base tracking principle. A heuristic algorithm is then used at the receiving end to design the digital synthesizer and the analog synthesizer. The simulation results show that the proposed hybrid precoding scheme has lower error rate than the traditional singular value based hybrid precoding, and the spectral efficiency is closer to the ideal value.