Tensor decomposition estimator applied to coherent targets in EMVS-MIMO radar

作     者：Yu, Zhiyu Chen, Qiulin Wen, Fangqing Li, Zhengzhou 

作者机构：Chongqing Univ Sch Microelect & Commun Engn Shazheng St 174 Chongqing 400044 Peoples R China Hubei Univ Automot Technol Inst Vehicle Informat Control & Network Technol 167 Checheng West Rd Shiyan 442002 Hubei Peoples R China China Three Gorges Univ Hubei Key Lab Intelligent Vis Based Monitoring Hyd 8 Daxue Rd Yichang 443002 Hubei Peoples R China 

出 版 物：《PHYSICAL COMMUNICATION》 (Phys. Commun.)

年 卷 期：2025年第69卷

核心收录：

学科分类：0810[工学-信息与通信工程] 0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Electromagnetic vector sensor Direction of arrival estimation PARAFAC decomposition Coherent targets MIMO radar 

摘      要：Electromagnetic vector sensor (EMVS) arrays in multiple-input multiple-output (MIMO) radar systems have gained substantial attention for their enhanced capabilities in target detection and localization over the past decade. However, conventional direction of arrival (DOA) estimation algorithms exhibit significant limitations when processing coherent targets. In response to these challenges, this paper introduces a novel algorithm leveraging tensor decomposition to enhance DOA estimation accuracy within monostatic EMVS-MIMO radar systems. Initially, we employ a specific sequence to rearrange the original array output, thereby linking the source matrix with the spatial response of the transmitting or receiving array. This rearrangement effectively addresses the rank deficiency issue in the covariance matrix. Subsequently, the restructured model undergoes parallel factor (PARAFAC) decomposition to yield high-precision estimates of the factor matrices. Ultimately, 2D-DOA estimation is accomplished by applying the normalized vector cross product (NVCP) technique to the polarization response factor matrices. In contrast to existing algorithms for coherent targets estimation, the proposed method ensures no information loss, achieves superior angle estimation accuracy, and is applicable to arbitrary sensor geometries. Numerical simulations substantiate the effectiveness and superiority of the proposed algorithm.