基于CSINN的相控阵天线波束赋形研究

Research on Beamforming of Phased Array Antenna Based on CSINN

混合波束赋形技术能在较少的性能损失前提下大幅降低系统的复杂度、能耗和成本。基于Saleh-Valenzuela的毫米波信道模型,针对部分信道状态信息(CSI)已知的情形,提出一种基于信道状态信息神经网络(CSINN)的相控阵天线波束赋形方法。该方法克服单纯模拟波束赋形的硬件约束,经过大量数据的多次迭代,CSINN不断学习毫米波复杂的信道传输特性。与基于流形优化的传统混合波束赋形算法相比,实验结果表明该方法具有明显的优势,有较强的鲁棒性和较低的复杂度,提升系统性能。

Hybrid beamforming technology can greatly reduce system complexity,energy consumption,and cost with less performance loss.Based on Saleh-Valenzuela's millimeter-wave channel model,this paper proposes a phased array antenna beamforming method based on the Chan⁃nel State Information Neural Network(CSINN)for situations where some Channel State Information(CSI)is known.This method overcomes the hardware constraints of pure analog beamforming.After multiple iterations of large amounts of data,CSINN continuously learns the com⁃plex channel transmission characteristics of millimeter waves.Compared with the traditional hybrid beamforming algorithm based on mani⁃fold optimization,the experimental results show that the method in this paper has obvious advantages,has stronger robustness and lower complexity,and improves system performance.