基于矩阵填充的子阵重构二维波达方向估计算法  

Subarray reconstruction 2D-DOA estimation algorithm based on matrix completion

作　　者：曾文浩[1] 朱晓华[1] 李洪涛[1] 庄珊娜[2] 

机构地区：[1]南京理工大学电子工程与光电技术学院,江苏南京210094 [2]石家庄铁道大学信息科学与技术学院,河北石家庄050043

出　　处：《南京理工大学学报》2017年第3期337-343,共7页Journal of Nanjing University of Science and Technology

摘　　要：为提高稀疏阵列下二维波达方向(2D-DOA)估计的效率,提出1种基于加速近邻梯度矩阵填充的子阵重构旋转子空间(APG-SRESPRIT)算法。建立了基于矩阵填充的稀疏阵列DOA估计信号模型,并验证该信号模型满足零空间性质。通过加速近邻梯度算法将该信号模型恢复为完整信号,划分子阵并构建合并矩阵。对合并矩阵进行奇异值分解,在子阵重构后估计目标角度,且目标角度自动配对。仿真实验表明该文算法可减少70%的阵元数量,且在稀疏阵列下准确估计2D-DOA。An accelerated proximal gradient singular value thresholding based subarray reconstruct ESPRIT （ APG- SRESPRIT） algorithm is proposed to improve the efficiency of two-dimensional direction- of-arrival （ 2D-D0A） estimation of sparse arrays. A DO A estimation signal model of sparse arrays is built based on matrix completion, and is proved to meet the null space property （ NSP）. The model is recovered to a complete signal model via accelerated proximal gradient singular value thresholding（ APG） , and subarrays are reconstructed to build a merged matrix. The singular value decomposition （ SYD） of the merged matrix is solved, the target angles are obtained after subarray re-construction, and the target angles can match automatically. Simulation experiments show that this algorithm can decrease the array number by 7 0 % ,and estimate the 2D-DOA of sparse arrays precisely.

关 键 词：阵列信号处理 矩阵填充 子阵重构 波达方向估计 稀疏阵列 零空间性质 奇异值分解 

分 类 号：TN911.72[电子电信—通信与信息系统]