塔克张量域正则极化ESPRIT

发布时间：2018-12-08 19:57

【摘要】：针对极化敏感阵列信号波达方向(direction of arrival,DOA)估计问题,提出了一种基于塔克张量域序贯截断高阶奇异值分解的正则极化旋转不变参数估计(Tucker tensor based regularized polarimetric estimation of signal parameters via rotational invariance technique,trpESPRIT)方法。首先对阵列接收信号进行塔克张量建模,之后通过序贯截断高阶奇异值分解获得塔克张量域信号子空间,最后利用多旋转不变子空间幅相关系获得信号DOA估计。相比于传统矩阵建模方法,塔克张量建模更便于组织多维数据结构,实现高维的数据匹配操作,而序贯截断高阶奇异值分解则可以获得更高的信号子空间估计精度以及后续的DOA估计。仿真结果表明,trpESPRIT方法较之常规矩阵方法和矢量方法可以更好地抑制噪声,具有更高的信号DOA估计精度,在低信噪比和低快拍条件下仍然具有良好的分辨能力。
[Abstract]:To solve the problem of DOA (direction of arrival,DOA) estimation of polarization-sensitive array signals, a new method for estimating the regular polarization rotation invariant parameters (Tucker tensor based regularized polarimetric estimation of signal parameters via rotational invariance technique,) based on Tucker tensor domain sequential truncation high order singular value decomposition is proposed. TrpESPRIT) method. The array received signal is modeled by Tucker Zhang Liang, then the signal subspace in Tucker tensor domain is obtained by sequential truncation of higher order singular value decomposition. Finally, the signal DOA estimation is obtained by using the amplitude-phase relation of multi-rotation invariant subspace. Compared with the traditional matrix modeling method, Tucker Zhang Liang modeling is more convenient to organize multidimensional data structure and achieve high-dimensional data matching operation. Sequential truncated higher-order singular value decomposition can obtain higher signal subspace estimation accuracy and subsequent DOA estimation. The simulation results show that the trpESPRIT method can suppress noise better than the conventional matrix method and vector method, and has higher signal DOA estimation accuracy, and still has a good resolution under low SNR and low rapid-shoot conditions.
【作者单位】：北京理工大学信息与电子学院;
【基金】：国家自然科学基金(61490691,61331019)资助课题
【分类号】：TN911.7

【相似文献】