This letter considers two-dimensional direction of arrival (DOA) estimation of coherent signals exploiting a moving uniform rectangular array. The motion of the array induces phase variations in the received signals across spatial positions, enabling the construction of decorrelated covariance matrices through forward-backward spatial smoothing. We analyze the achievable degrees of freedom (DOFs) in terms of movement steps and examine the impact of the motion support on effective decorrelation. Notably, we show that the maximum number of DOFs can be achieved if each movement step is at least half the signal wavelength and the number of movement steps is no less than half the number of array elements. Furthermore, it is demonstrated that distributing motion across both array axes yields better decorrelation and estimation performance than restricting movement to a single dimension.

Hybrid analog-digital beamforming is an effective approach for practical implementations of a massive multiple-input multiple-output (MIMO) system by reducing the number of radio frequency (RF) chains. Fully connected hybrid beam-forming (F-HBF), where each RF chain is connected to each antenna, can lower hardware complexity, power consumption, and cost compared to digital beamforming. Subarray-based hybrid beamforming (S-HBF), where a specific group of RF chains is allocated to a particular subarray, can further reduce hardware requirements. The antenna array is divided into subarrays using effective partitioning so that the optimization of analog beamforming can be shared across multiple subarrays, substantially reducing computational complexity.

This paper addresses the direction-of-arrival (DOA) estimation problem of mixed coherent and uncorrelated signals using a sparse rectangular array, where tensor reconstruction is employed to preserve the structure of multi-dimensional array signals. In the proposed approach, we first estimate the DOAs of uncorrelated signals using the subspace algorithm. After eliminating the contribution of uncorrelated signals from the covariance tensor, a structural tensor decorrelation process is introduced to decorrelate the resulting coherent covariance tensor. The canonical polyadic decomposition method is employed to the decorrelated covariance tensor to detect the coherent signals. The conditions of signal resolvability are analyzed.

This letter develops a new direction-of-arrival (DOA) estimation method for mixed coherent and uncorrelated signals through the reconstruction of a set of Toeplitz matrices. More specifically, Toeplitz matrices are formed by utilizing the rows and columns of the covariance matrix, and their average is used by subspace-based algorithms to effectively estimate the signal DOAs. Compared to existing methods, the proposed approach provides a high number of degrees of freedom and requires a low computation complexity.

This paper considers a wireless network assisted by an intelligent reflecting surface (IRS) to enhance data transmission between the base station and mobile users. Our objective is to estimate and predict the user-IRS channels by exploiting a small number of sparsely distributed active elements with a low pilot overhead. The Hermitian and Toeplitz properties of the data covariance matrices are used to perform covariance matrix interpolation for enhanced estimation of the time-varying user-IRS multipath channels, and a machine learning-based channel predictor is developed to predict the channels based on prior channel estimates so as to shorten the required training pilot signals and enhance the transmission data rate. Simulation results verify the effectiveness of the proposed method for accurate channel estimation and prediction.

In this paper, we develop a deep learning framework to optimize the compressive sampling matrix in a massive multiple-input multiple-output (MIMO) system. The optimized compressive sampling matrix is utilized to project high-dimensional data received at the massive MIMO system into a lower-dimensional space so that the directions of arrival and other signal parameters can be efficiently obtained with a reduced hardware complexity. The proposed deep learning approach for optimizing the compressive measurement matrix increases its robustness and generalizability.

Distributed array consisting of multiple subarrays is attractive for high-resolution direction-of-arrival (DOA) estimation when a large-scale array is infeasible. To achieve effective distributed DOA estimation, it is required to transmit information observed at the subarrays to the fusion center, where DOA estimation is performed. For noncoherent data fusion, the covariance matrices are used for subarray fusion. To address the complexity involved with the large array size, we propose a compression framework consisting of multiple parallel encoders and a classifier. The parallel encoders at the distributed subarrays are trained to compress the respective covariance matrices. The compressed results are sent to the fusion center where the signal DOAs are estimated using a classifier based on the compressed covariance matrices.