Multi-SVD based subspace estimation to improve angle estimation accuracy in bistatic MIMO radar

A new multi-SVD based subspace estimation algorithm is proposed to improve the direction of departure (DOD) and direction of arrival (DOA) estimation accuracy for bistatic multiple-input multiple-output (MIMO) radar. First, the matched filter output is transformed to a 3-order tensor. Then the signal subspace is estimated by multi-SVD of the matrix unfoldings of this tensor or its covariance tensor. Since the multidimensional structure is utilized, the estimated signal subspace of our method has better accuracy than that of the traditional SVD/EVD method. Combined with the classical subspace based methods, such as MUSIC and ESPRIT, the proposed approach improves the angle estimation performance compared with the existing ones. Simulation results confirm the effectiveness of our method.

► The data is modeled by a 3-order tensor, and the covariance tensor is defined.
► The noise is filtered out in each mode of the above two tensors.
► Multi-SVD yields more accurate signal subspace estimation than classical methods.
► Multi-SVD is suitable for any subspace based method to improve DOD-DOA estimation.
► The analytical performance of multi-SVD method is given.