Abstract: Based on L-shaped linear array, this paper presents a fast method for the two-dimensional （2D） Direction-Of-
Arrival（DOA） estimation of Coherently Distributed（CD） source. The presented method obtains two rotational invariance matrixes about the central DOAs of CD sources by one order Taylor approximation to the generalized steering vectors of two pairs of shifted subarrays. The central DOAs of 2D CD sources are estimated using a new propagator method to solve the rotational invariance matrixes. The method avoids spectrum-peak searching and the eigen decomposition of the high-dimensional sample covariance matrix in classical subspace methods. As a result, the computational cost is significantly reduced. Under small angular spread, the proposed method provides a good estimation performance and outperforms Sequential One-dimensional Searching（SOS） algorithm at low SNR. As the new method is a blind estimator, the prior information of the specific angular distribution shapes of distributed sources is not necessary. Simulation results demonstrate the effectiveness of the method.

Key words: coherently distributed sources, 2-dimensional Direction-Of-Arrival（DOA） estimation, L-shaped linear array, rotational invariance matrix, propagator method

摘要： 基于L型线阵，提出了一种估计相干分布源二维波达方向（DOA）的快速算法。通过对两组平移子阵的广义方向矢量做泰勒近似获得关于分布源中心DOA的两个旋转不变矩阵，利用传播算子法求解旋转不变矩阵从而估计出分布源的中心DOA。该算法避免了常规子空间算法中的谱峰搜索和对高维样本协方差矩阵做特征分解，显著降低了计算量。算法在小角度扩展情形下有优异的估计性能，且低信噪比时的估计性能优于一维交替搜索算法。此外，算法无需知道分布源的角分布形式，是一种盲估计。仿真结果验证了算法的有效性。

关键词: 相干分布源, 二维波达方向估计, L型线阵, 旋转不变矩阵, 传播算子法