Objective To address the problems of structural complexity, large size, and high cost in traditional direction-finding platforms and to achieve miniaturization and portability of direction-finding equipment, this paper proposes a multi-snapshot atomic norm minimization direction-of-arrival (DOA) estimation algorithm based on channel switching.

Method The proposed approach exploits a channel switching mechanism in which a switching matrix is designed to randomly vary across sampling instances. This strategy enables a reduced number of radio-frequency (RF) channels to sequentially sample a larger antenna array, thereby effectively emulating a virtual array aperture while significantly lowering hardware requirements. Based on the switched observations, a noisy multi-snapshot signal model is established. The DOA estimation task is then formulated as a continuous-domain atomic norm minimization problem, which avoids the basis mismatch issue commonly encountered in grid-based sparse reconstruction methods. By solving the resulting semidefinite programming (SDP) problem, a structured Toeplitz covariance matrix is recovered. The DOA parameters are subsequently extracted through Vandermonde decomposition of this Toeplitz matrix, yielding high-resolution angle estimates. In addition, to provide a theoretical benchmark for performance evaluation, the Cramér–Rao bound (CRB) under the proposed channel switching observation model is rigorously derived.

Results Extensive numerical simulations are conducted to assess the effectiveness of the proposed method. The results indicate that, with 24 antenna elements, 12 RF channels, a signal-to-noise ratio (SNR) of 5 dB, and 100 snapshots, the proposed algorithm achieves a root mean square error (RMSE) of less than 0.1°. Furthermore, when the angular separation between two closely spaced sources is as small as 0.6°, the proposed method attains a resolution success probability of 100%, demonstrating its strong super-resolution capability. Compared with conventional sparse reconstruction–based algorithms such as orthogonal matching pursuit (OMP) and classical subspace-based methods such as MUSIC, the proposed approach exhibits significantly improved estimation accuracy and angular resolution. Moreover, as the SNR and the number of snapshots increase, the estimation performance of the proposed algorithm progressively approaches the derived CRB, indicating near-optimal efficiency.

Conclusion The proposed method effectively reduces system complexity and hardware cost while maintaining high direction-finding accuracy and angular resolution, providing a feasible solution for compact and high-precision DOA measurement systems.