Investigo. Repositorio Institucional de la Universidade de Vigo
Publicación
Artículo científico (JournalArticle)
Generalized least squares based covariance reconstruction for DoA estimation with hybrid partially connected arrays
Accurate Direction of Arrival (DoA) estimation with
Partially-Connected (PC) arrays is hindered by dimensionality compression in the analog domain. This loss of spatial information restricts the degrees of freedom (DoF) available, degrading
the performance of high-resolution algorithms. To recover the lost
DoFs, we propose an approach that reconstructs the equivalent
spatial covariance matrix (SCM) of the fully digital array. The
method leverages the discrete Short-Time Fourier Transform
(STFT) underlying the PC measurements, and formulates SCM
reconstruction as a Generalized Least Squares (GLS) estimation
problem. The resulting estimator, asymptotically efficient, is
validated across multiple scenarios with numerical results.
Index Terms—Direction of arrival estimation, hybrid analog
and digital arrays, covariance matrix reconstruction, array signal
processing, Agencia Estatal de Investigación | Ref. PID2021-122483OA-I00, Agencia Estatal de Investigación | Ref. PID2022-136512OB-C22, Xunta de Galicia | Ref. ED481A 2022/407