Robust smoothed ℓ0-norm based approach for MIMO radar target estimation

摘要

To enable the smoothed ℓ0-norm (SL0) algorithm to offer accurate estimates of the target parameters in three dimensions of range, angle, and Doppler, the fine discretisation of the potential target space is required for multiple-input multiple-output (MIMO) radars, which results in the ill-conditioned sensing matrix. Unfortunately, the SL0 algorithm will provide unacceptable results since the large errors occur in computing initial value and performing projection onto the feasible set through the use of the pseudo-inverse of the ill-conditioned sensing matrix. In this study, the authors present a robust SL0 approach for MIMO radars to provide accurate angle-range-Doppler estimates. The appropriate permutation matrix, which takes the place of the pseudo-inverse of the sensing matrix in implementing SL0 algorithm, can be pre-computed by taking advantage of the bi-conjugate gradient stabilised approach and the singular value decomposition (SVD) of the sensing matrix. Simulation results show that the proposed algorithm not only has lower computational cost, but provides better performance in estimation of range-angle-Doppler parameters, compared with the regularised iterative reweighted minimisation approach, SL0 algorithm and its modified versions in combination with Tikhonov and truncated SVD methods.