一种基于平行因子分析的近场源定位新方法

A New Near-Field Source Localization Algorithm Using Parallel Factor Analysis

本文提出了一种基于平行因子分析的近场窄带信源频率、距离及二维到达角联合估计新算法.首先给出一种新的十字阵列;然后结合Toeplitz矩阵的结构特点,选择特定序号阵元输出计算的高阶累积量巧妙构造5个Toeplitz矩阵;接着在高阶累积量域构造平行因子分析模型,分析了该模型低秩分解的唯一性,并从其分解得到的矩阵中联合估计信源参数.与现有方法相比,该算法有效降低了阵列的孔径损失,无须参数配对.仿真结果表明该算法是有效的.

An estimation algorithm for frequency, range, and two-dimensional （2-D） direction-of-arrival （DOA） of each near-field narrowband source is presented in this paper. Firstly, we propose a new cross-array;secondly, we compute some cumulants using the properly chosen sensor outputs, and form five Toeplitz matrices by re-using these cumulants, and construct the parallel factor （PARAFAC） analysis model in the cumulant domain;thirdly, we analyze the uniqueness of the model＇s low-rank decomposition, and estimate the source parameters via the decomposition. In comparison with the conventional methods, the proposed method efficiently uses the array aperture, and doesn＇t require pairing parameters. Finally, the simulation results are presented to validate the performance of the proposed method.