Parameter Estimation With Bistatic MIMO Radar: A Coarray Tensor Decomposition Framework

The traditional parameter estimation method based on the matrix framework in multiple-input–multiple-output (MIMO) radar with sparse arrays loses the information of the tensor signal structure, resulting in performance degradation. Therefore, this article primarily investigates parameter estimation methods for bistatic MIMO radar based on a tensor decomposition framework. The conventional tensor based on self-correlation discards two virtual array elements, leading to a loss of performance and degrees of freedom (DOFs). To address this issue, this article proposes a coarray tensor decomposition framework for direction of departure (DOD) and direction of arrival (DOA) estimation, introducing an approach for constructing a coarray tensor. First, a virtual difference coarray is constructed using two subtensors. Then, a coarray tensor is constructed based on the non-Hermitian structure of the cross-correlation signal matrix. Next, the resulting coarray tensor is reconstructed to achieve optimal source identifiability. Nevertheless, the increase in dimensionality resulting from the reconstructed coarray tensor leads to higher algorithmic complexity. To mitigate this, we perform a real-valued transformation on the reconstructed coarray tensor, which speeds up the execution of the method. Additionally, the proposed method also features the capability to suppress colored noise. Theoretical analysis indicates that the reconstructed coarray tensor has more DOFs than the original coarray tensor. Simulation results show that the proposed method has larger DOFs and good parameter estimation performance.