中文摘要：

基于常见的具有平移不变子阵的平面传感器阵列,在子阵传感器任意拓扑布署下,列堆栈两个子阵的接收数据,进而构造分块相关矩阵。提出一种循环最大信号子空间估计方法,利用得到的信号子空间对分块相关矩阵进行降维,处理降维后的分块矩阵,在低秩空间重构出阵列导向矢量矩阵,估计出空间目标的二维到达方向（DOA）。计算机仿真表明：所提算法在信噪比、采样数变化下对目标的定位精度均优于精典2D ESPRIT算法。

英文摘要：

Based on the common planar sensor array with shift invariance subarrays, in the case of subarrays possess arbitrary topological structure, the received data of the subarrays were column stacked and the block correlation matrix was formed. A cyclic maximization method was proposed to estimate the signal subspace, the estimated signal subspace was then used to reduce the dimension of the block correlation matrix, and the steering vector matrix of sensor array was derived from the dimension reducing matrix, the two dimension direction of arrival （DOA） of spatial targets could be found from steering vector matrix. Simulation results show that the proposed algorithm possesses better location accuracy than the classical 2D ESPRIT algorithm.