Efficient DoA Estimation Algorithm with Arrays of Arbitrary Geometry via Guided Phase Manipulation

Exhaustive search-based Direction-of-Arrival (DoA) estimation algorithms are computationally intensive and may yield a biased result with arrays of arbitrary geometry. In this paper, an algorithm named Grid Iterative Estimation of Signal Parameters via Rotational Invariance Techniques (GI-ESPRIT) is proposed to get a statistically unbiased DoA estimate efficiently based on Augmented ESPRIT (AESPRIT). First, AESPRIT is modified with a loop structure, and then the DoA estimate is obtained via guided phase compensation iteratively. The iteration will converge provably to the right DoA value quickly if a proper initial compensation angle is provided. The entire angle domain is coarsely gridded and the direction of each grid is utilized as an initial compensation angle. Simulations verify the efficiency and effectiveness of the proposed method, and the computation efficiency can be improved significantly compared with other DoA methods such as MUltiple SIgnal Classification (MUSIC) and Manifold Separation Technique (MST).