电信科学 ›› 2019, Vol. 35 ›› Issue (6): 89-95.doi: 10.11959/j.issn.1000-0801.2019069

• 研究与开发 • 上一篇    下一篇

基于LFM的毫米波大规模MIMO混合预编码方法

陈浩椅,丁一凡,程英   

  1. 杭州电子科技大学通信工程学院,浙江 杭州 310016
  • 修回日期:2019-04-20 出版日期:2019-06-20 发布日期:2019-06-20
  • 作者简介:陈浩椅(1995- ),男,杭州电子科技大学通信工程学院硕士生,主要研究方向为毫米波通信。|丁一凡(1995- ),男,杭州电子科技大学通信工程学院硕士生,主要研究方向为无线通信。|程英(1995- ),女,杭州电子科技大学通信工程学院硕士生,主要研究方向为无线通信。

Hybrid precoding method for mmWave massive MIMO systems based on LFM

CHEN Haoyi,DING Yifan,CHENG Ying   

  1. School of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310016,China
  • Revised:2019-04-20 Online:2019-06-20 Published:2019-06-20

摘要:

模拟数字混合预编码是毫米波大规模多输入多输出(MIMO)系统的关键技术,可以减少硬件成本而又兼顾系统性能。但是传统的混合预编码方法往往需要寻找适合的码本进行预编码,有的码本实际情况中并不容易获得,或者有偏差。针对这个问题,提出一种无需码本的基于机器学习中隐因子模型(LFM)的模数混合预编码方法。利用LFM分解与随机梯度下降法,使设计的混合预编码矩阵逼近最优全数字预编码矩阵,从而得到极佳的性能。仿真结果表明,相比基于正交匹配追踪(OMP)算法的混合预编码设计方法,该方法不但无需事先设计码本,而且其性能比基于OMP算法的混合预编码方法更优秀,更接近最优的全数字预编码方法。

关键词: LFM, 毫米波, 大规模MIMO, 混合预编码

Abstract:

Analog-digital hybrid precoding is a key technology for millimeter wave massive MIMO systems that reduce hardware costs while balancing system performance.However,the traditional hybrid precoding scheme often needed to find a suitable codebook for precoding,and some codebooks were not easy to obtain or had deviations in actual situations.An analog-digital hybrid precoding method based on latent factor model (LFM) in machine learning without codebook was proposed for this problem.The LFM decomposition and stochastic gradient descent method were used to approximate the designed precoding matrix to the optimal full digital precoding matrix for good performance.The simulation results show that compared with the hybrid precoding design method based on orthogonal matching pursuit (OMP) algorithm,this method not only does not need a codebook,but also has better performance than the hybrid precoding algorithm based on OMP algorithm,which is closer to optimal full digital precoding method.

Key words: LFM, millimeter wave, massive MIMO, hybrid precoding

中图分类号: 

No Suggested Reading articles found!