FDA-MIMO Radar Parameter Estimation Based on Bayesian Tensor Decomposition

In practical applications, imperfections in antenna arrays and the unknown number of targets can degrade radar performance for target parameter estimation. This paper investigates the joint analysis of target count, range, and azimuth angle in FDA-MIMO radar systems, considering the electromagnetic coupling interference between array elements. An enhanced real-valued variational Bayesian CP decomposition algorithm is proposed based on traditional tensor decomposition theory to exploit the multidimensional features of received signals fully. First, a selection matrix is utilized to recover the Vandermonde structure of the array from the received signals. Then, within the variational Bayesian learning framework, the tensor rank and factor matrices of the multidimensional received tensor signal are jointly estimated, enabling target number and decoupled computation of angle-range parameters. Finally, numerical simulations validate the proposed algorithm's superiority in estimation accuracy.