应用双线性模型的离格波达角估计方法  

Off-grid DOA Estimation Based on Bi-linear Model

作　　者：王友顺 高童迪 袁正道 丁永春[3] WANG Youshun;GAO Tongdi;YUAN Zhengdao;DING Yongchun(Artificial Intelligence Engineering Research Center,Open University of Henan,Zhengzhou 450001,China;School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China;The 713th Research Institute of China State Shipbuilding Co.,Ltd.,Zhengzhou 450001,China)

机构地区：[1]河南开放大学人工智能工程研究中心,郑州450002 [2]郑州大学信息工程学院,郑州450001 [3]中国船舶集团有限公司第七一三研究所,郑州450001

出　　处：《电讯技术》2022年第4期482-488,共7页Telecommunication Engineering

基　　金：中国博士后科学基金面上项目(2019M652576);国家自然科学基金资助项目(61901417);河南省科技攻关项目(222102210328,222102210184,212102210421,212102210542);河南省高等学校重点研究课题(20B510005)。

摘　　要：针对离格波达角估计(Direction of Arrival,DOA)问题,提出了一种基于双线性向量化近似消息传递(Bilinear Vectorization Approximate Message Passing,BAd-VAMP)的解决方法。该方法首先将DOA估计问题建模为双线性模型,利用BAd-VAMP算法进行迭代估计;其次通过误差向量和来波信号的相同稀疏特征,缩减矩阵维度,以降低迭代算法的复杂度;最后提出一种基于残差的冷/热重启机制,避免BAd-VAMP算法陷入鞍点,提升算法的鲁棒性。仿真结果表明,相比已有算法,所提方法在快拍次数较少的情况下具有显著的性能增益,使其更适于机载雷达等快时变场景。A method based on bilinear vectorization approximate message passing(BAd-VAMP)is proposed to estimate the off-grid direction of arrival(DOA)problem.Firstly,the DOA model is modeled as a bilinear model,and the BAd-VAMP algorithm is used for iterative estimation.Secondly,the dimension of the observation matrix is reduced by the same sparse feature of the error vector and incoming signal to reduce the complexity of the iterative algorithm.Finally,a cold/warm restart mechanism based on residual error is proposed to avoid the saddle point of BAd-VAMP algorithm and improve the robustness.The simulation results show that compared with the existing algorithms,the proposed method has tremendous performance gain in the case of fewer snapshots,making it more suitable for fast time-varying scenes such as airborne radar.

关 键 词：离格波达角估计 双线性模型 泰勒展开 稀疏特性 迭代阻尼 

分 类 号：TN911.7[电子电信—通信与信息系统] TN971[电子电信—信息与通信工程]