Abstract:

To address the challenge of underwater array direction of arrival (DOA) estimation of poor discrimination between adjacent sound sources when only a few snapshots are available, a fast estimation method based on iterative atomic norm minimization (IANM) is proposed for estimating the DOA for uniform circular arrays. To reduce the impact of non-uniform noise caused by modal domain processing, the proposed method preprocesses the array manifolds using the modal domain processing technique. This technique transforms the uniform circular array (UCA) into a virtual linear array and then estimates the covariance matrix of the received signal in the absence of noise through diagonal reconstruction. In order to make full use of the sparsity of the received signal and avoid the error caused by dictionary grid search, an IANM sparse recovery method is introduced in the modal domain, and the UCA-IANM direction estimation method is proposed. The sparse recovery problem of atomic norm minimization (ANM) is typically addressed using the interior point method, which leads to a sharp increase in workload as the number of received signal snapshots grows, making it unsuitable for scenarios with limited underwater computational resources. In this paper, based on the alternating direction multiplier method (ADMM), the UCA-IANM assisted by ADMM with parameter optimization (UCA-IANM-APO) DOA fast estimation method is proposed to address the challenge in selecting the regularization parameter. Simulation experiments and measured data demonstrate that the UCA-IANM-APO method has better angular resolution and estimation accuracy compared to the traditional DOA estimation methods, and the solution speed is improved by two orders of magnitude compared to the interior point method.

Key words: uniform circular array (UCA), virtual array, diagonal reconstruction, atomic norm minimization (ANM), alternating direction multiplier method