基于非圆信号特征的实值张量ESPRIT算法

基于最大非圆率非圆信号特点,提出一种实值张量旋转不变子空间(estimation signal parameters via rotational invariance techniques, ESPRIT)算法。首先,通过研究张量与矩阵之间的转化关系,将阵列接收数据矩阵推广到张量空间;然后,利用欧拉公式将阵列接收数据张量转化成余弦与正弦数据张量,根据阵列维数将其分别在各维上加以拼接,并对拼接的实值数据张量做高阶奇异值分解,获取信号子空间;最后,通过构造选择矩阵和进行特征分解,来联合估计阵列各维相位差,实现波达方向估计。实验仿真结果表明,此算法具有良好的分辨力和测角精度。

Real-value tensor ESPRIT algorithm based on characteristics of non-circular signals

Based on the characteristics of non-circular signals of the maximum non-circular rate, a real-value tensor estimation signal parameters via rotational invariance techniques (ESPRIT) algorithm is put forward. Firstly, the array receiving data matrix is extended to tensor space by studying the relationship between tensor and matrix. Then, the array receiving data tensor is turned into cosine and sine data tensor by using the Euler formula, according to the array dimensions to concatenate cosine and sine data tensor in each dimension, respectively, and obtain the signal subspace through a higher-order singular value decomposition of real-value concatenation tensor data. Finally, phase difference is estimated jointly in each dimension to realize direction of arrival (DOA) estimation by constructing the selection matrix and eigen decomposition. The experimental simulation results show that this algorithm has a better resolution and angle measurement accuracy.