RM–OMP Scheme for RIS‐Assisted Single‐Antenna DOA Estimation

ABSTRACT

In reconfigurable intelligent surface (RIS)‐assisted single‐antenna direction‐of‐arrival (DOA) estimation, existing low‐overhead methods often suffer from degraded accuracy under limited observations and from the high complexity of accurate off‐grid refinement. To address these issues, this letter proposes a low‐overhead DOA estimation scheme termed residual‐minimisation (RM)–Orthogonal Matching Pursuit (OMP). First, a phase‐compensated RIS codebook is designed to suppress the RIS‐to‐AP channel effect and simplify the sensing model. Second, atom selection is performed by minimising the residual, which makes the scheme more robust to sensing‐matrix coherence under severe compression. Third, a low‐complexity closed‐form off‐grid correction method is developed to mitigate off‐grid bias without iterative refinement. Simulation results show that RM–OMP enables reliable DOA estimation with as few as 8 RIS state changes, significantly outperforms conventional OMP‐based schemes under limited switching, and approaches the accuracy of NOMP when more observations are available.