Abstract
A lot of effort has been made recently to increase the angular performance of automotive radar sensors. At the same time, low system costs are still favored so that technical solutions like multiple-input and multiple-output (MIMO) and sparse arrays have found their way into the market successfully. With these techniques, however, some tradeoffs regarding sidelobe level and ambiguity are inevitable which impose new challenges to angle estimation methods. This paper presents a novel Fast Variational Bayesian (FVB) based direction of arrival (DoA) estimator suitable for mitigating the effects of high sidelobes in sparse arrays. The proposed algorithm is firstly adapted to automotive MIMO radar. Super-resolution and multi-target capability are validated by extensive experimental evaluations based on synthetic and measured radar data. The presented approach performs best in separating closely spaced reflections amongst all other accelerated Sparse Bayesian algorithms reported in literature so far. Furthermore, it is shown that FVB can outperform other state-of-the-art algorithms like beamforming or maximum likelihood methods.
Users
Please
log in to take part in the discussion (add own reviews or comments).