Recurrent Design of Probing Waveform for Sparse Bayesian Learning Based DOA Estimation

Direction-of-arrival (DOA) estimation can be represented as a sparse signal recovery problem and effectively solved by sparse Bayesian learning (SBL). For the DOA estimation in active sensing, the SBL-based estimation error is related to the transmitted probing waveform. Therefore, it is expected to improve the estimation by waveform optimization. In this paper, we propose a recurrent scheme of waveform design by sequentially leveraging on the previous-round SBL estimates. Within this scheme, we formulate the waveform design problem as a minimization of the SBL estimation variance, which is nonconvex and then solved by a majorization-minimization based algorithm. The simulations demonstrate the efficacy of the proposed design scheme in terms of avoiding incorrect detection and accelerating the DOA estimation convergence. Further, the results indicate that the waveform design is essentially a beampattern shaping methodology.