A Joint Range-Azimuth Off-Grid Recovery Algorithm for FMCW Automotive Radar

Obtaining an accurate range-azimuth map is very important in automotive radar applications. Since predefined grids cannot exactly match true values, the range-azimuth map obtained by traditional signal processing has bias. This paper proposes a joint algorithm to achieve gridless range-azimuth map recovery in frequency modulated continuous wave (FMCW) radar. First, we present the FMCW radar signal model, and show the limitation of grids in signal recovery. Next, we describe the proposed joint range-azimuth (JRA) algorithm, and discuss the convergence guarantee. Moreover, considering real-time processing in automotive radar, an acceleration module (AM) is designed to reduce the running time of JRA. AM utilizes swarm information to simultaneously recover multiple scattering points. Lastly, numerical simulations confirm that JRA outperforms state-of-the-art off-grid/gridless recovery algorithms in multiple aspects.