Fast iterative interpolated beamforming for high fidelity single snapshot DOA estimation

Abstract

We propose fast direction-of-arrival (DOA) estimation of multiple sources using a single snapshot. This problem arises in many practical applications and is pertinent to automotive radar. Existing beamforming algorithms incur a high computational cost as they usually require significant zero-padding or fail to fully remove the bias that results from the spectral leakage, especially at high signal to noise ratios (SNRs). We demonstrate superior performance of an iterative interpolated beamforming algorithm which is based on a recently introduced fast and accurate frequency estimation method. This algorithm has a computational complexity of the same order as the fast Fourier transform (FFT), yet is capable of delivering high-fidelity multi-source DOA estimates that can achieve the CRB. Two modifications are proposed to further enhance the algorithm threshold performance and reduce its computational cost. DOA estimation accuracy and computational simplicity of the proposed technique are demonstrated using simulations involving relevant automotive scenarios.