OPTIMIZATION OF ESPRIT ALGORITHM AND FAST DOA ESTIMATION FOR UNIFORM LINEAR ARRAYS

To tackle the problems of high computational complexity and cumbersome engineering deployment posed by traditional subspace-based direction-of-arrival (DOA) estimation algorithms, this study presents a low-complexity DOA estimation approach designed specifically for uniform linear arrays (ULA). The key innovation of the proposed method is the elimination of eigenvalue decomposition or singular value decomposition (EVD/SVD)—two core components of classical subspace-based techniques. Instead, it employs a strategy of array segmentation and matrix reconstruction. In detail, the ULA is first split into two separate subarrays, and their cross-covariance matrix is used to build a joint cross-covariance matrix. An equivalent signal subspace is then directly generated via a linear transformation, which greatly simplifies the computational procedure. Finally, by exploiting the rotation invariance principle of ESPRIT, the proposed approach realizes fast and precise DOA estimation. Theoretical analyses and simulation results indicate that, in comparison with traditional subspace-based algorithms, the proposed method reduces computational complexity significantly while retaining comparable estimation accuracy, thus improving the overall efficiency of DOA estimation.