Spatial Angular Pseudo-Derivative Search Algorithm: A Real-Time Single-Snapshot Super-Resolution Sparse DOA Scheme for Automotive Radar

Abstract

Accurate, high-resolution, and real-time DOA estimation plays a crucial role in automotive radar perception. While sparse signal recovery techniques offer super-resolution and high-precision estimation, their prohibitive computational complexity remains a primary bottleneck for practical deployment. This paper proposes a sparse DOA estimation scheme specifically tailored for the stringent requirements of automotive radar such as limited computational resources, restricted array apertures, and single-snapshot constraints. By leveraging the spatial angular pseudo-derivative (SAPD) property of the sparse DOA solutions and incorporating this property as a constraint into an $\ell_0$-norm minimization problem, the proposed formulation transforms unordered exhaustive verification of candidate solutions into an ordered search process for sparse DOA estimation. Thus, the associated solver, called the SAPD Search algorithm, naturally transforms the high-dimensional optimization task into an efficient grid-search scheme. The SAPD search algorithm circumvents high-order matrix inversions and computationally intensive iterations. We also provide an analysis of the computational complexity of the proposed algorithm. Numerical simulations and experimental validation demonstrate that the SAPD Search algorithm achieves a superior balance of millisecond-level computational efficiency, high precision, and super-resolution, making it highly suitable for next-generation automotive radar applications.