基于深度学习的波达方向估计方法综述  

Survey of DOA Estimation Methods Based on Deep Learning

作　　者：刘其有 何瑞 施伟 LIU Qi-you;HE Rui;SHI Wei(School of Electronic and Information Engineering,Nanjing University of Information Science&Technology,Nanjing 210044,China;The Sixty-third Research Institute,National University of Defense Technology,Nanjing 210007,China)

机构地区：[1]南京信息工程大学电子与信息工程学院,江苏南京210044 [2]国防科技大学第六十三研究所,江苏南京210007

出　　处：《中国电子科学研究院学报》2025年第1期1-9,共9页Journal of China Academy of Electronics and Information Technology

基　　金：国家自然科学基金资助项目(62401593)。

摘　　要：在阵列信号处理中,波达方向(Direction of Arrival,DOA)估计是一项核心任务,旨在确定信号源相对于阵列传感器的方向。随着深度学习(Deep Learning,DL)技术的兴起,其在DOA估计领域的应用已取得显著成果。本研究回顾了22篇关于基于DL的DOA估计方法的文献,分析了这些研究的发表背景、引用频次,并系统探讨了DL在DOA估计中的具体应用,包括目标、模型和方案。通过评价指标(如RMSE、MSE)对这些方法进行了评估,并探讨了信噪比、信号数量、阵元数和快照数等因素对DOA估计的影响。此外,本研究还特别关注了多输入多输出(Multiple Input Multiple Output,MIMO)雷达阵列在多源DOA估计方面的最新进展。Direction of Arrival(DOA)estimation,pivotal in array signal processing,determines signal source directions relative to array sensors.The integration of deep learning(DL)has significantly advanced DOA estimation.This review examines 22 DL-based DOA estimation approaches,assessing their publication context and scholarly impact.It delves into DL’s role,encompassing objectives,models,and schemes,and evaluates them using metrics like RMSE and MSE.The review also investigates factors affecting DOA estimation,such as signal-to-noise ratio(SNR),signal count,array element number,and snapshot quantity.Special emphasis is placed on advancements in multi-source DOA estimation for MIMO radar arrays.

关 键 词：深度学习 DOA估计 阵列误差 MIMO雷达 

分 类 号：TN957[电子电信—信号与信息处理]