基于近似l0范数的实数化DOA估计  

REAL-VALUED DOA ESTIMATION BASED ON APPROXIMATE l0 NORM

作　　者：吉爱国 刘伟平 刘志强 Ji Aiguo;Liu Weiping;Liu Zhiqiang(School of Information and Control Engineering,Qingdao University of Technology,Qingdao 266520,Shandong,China)

机构地区：[1]青岛理工大学信息与控制工程学院,山东青岛266520

出　　处：《计算机应用与软件》2020年第11期328-333,共6页Computer Applications and Software

摘　　要：针对利用压缩感知进行波达方向(Direction of Arrival,DOA)估计时求解l_0范数NP难、噪声敏感等问题,提出一种基于近似l_0范数的实数化DOA估计算法(AL0-DOA)。对阵列接收数据的协方差矩阵进行Khatri-Rao(KR)积变换,将阵列多测量矢量模型转换为虚拟阵列单测量矢量模型,并通过降维和实数化进一步降低计算量,同时抑制噪声,提高DOA估计的准确性。利用信源在空间的稀疏性构造冗余字典,引入平滑函数来近似l_0范数,将无法直接求解l_0范数问题转化为平滑函数的最优化问题,可通过修正牛顿算法快速求解。仿真结果表明该算法计算快,精度较高,可对DOA进行有效估计。In order to solve the problem that l0 norm is NP-hard and noise-sensitive when using compressed sensing to estimate direction of arrival(DOA),a real-valued DOA estimation algorithm using approximate l0 norm(AL0-DOA)is proposed.Khatri-Rao(KR)product transformation was applied to the covariance matrix of the received data,and the array multiple measurement vector model was converted into a virtual array single measurement vector model.Dimensionality reduction and realization were used to reduce the computational complexity,suppress noise and improve the accuracy of DOA estimation.Redundant dictionary was constructed by the sparseness of sources in spatial.The smoothing function was introduced to approximate l0 norm,which could not directly solve the l0 norm problem into the optimization problem of smoothing function,which could be solved quickly by modified Newton algorithm.The simulation results show that the proposed algorithm is fast in calculation and high in accuracy,and can effectively estimate DOA.

关 键 词：波达方向估计 近似l_0范数 实数化 KHATRI-RAO积 压缩感知 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]