In this article, we focus on indoor direct positioning using a practical massive multiple-input and multiple-output (MIMO) array with imperfections. First, the imperfect array needs to be calibrated in an anechoic chamber from the view of array signal processing. This brings inconveniences due to the large size of massive MIMO systems. To solve this problem, we propose onsite calibration, where a banded calibration matrix, calibrating both phase-gain error and mutual coupling, can be estimated by performing least-squares on the measure near-field channels, i.e., the site survey data. The feasibility of this calibration method can be explained by that the non-line-of-sight (NLOS) channel is able to be modeled using Rayleigh fading; i.e., the sum of the multipath components can be treated as Gaussian noise. Then, to estimate the user position, we propose a new two-stage direct positioning scheme. In the first stage, the coarse position is obtained via global search-based conventional beamforming (CBF) using a global calibration matrix. In the second stage, fine positioning is achieved via local search-based multiple signal classification (MUSIC) using a local calibration matrix. The proposed method is computationally efficient, insensitive to model order selection, free of dense site survey, and robust to partial line-of-sight (LOS) blocking. Finally, the proposed method is verified by practical measurements from a distributed array (DA), uniform linear array (ULA), and uniform rectangular array (URA), and compared with a deep learning positioning-method. The results show that the DA with the proposed method performs the best, achieving sub-centimeter accuracy (typically below 5 mm).

Direction-of-arrival (DOA) estimation under low SNR and snapshot deficient scenarios has practical application value. Since the array output of the uniform linear array (ULA) can be viewed as a linear combination of the equal interval sampling complex exponentials (EISCE), it can be annihilated by a spatial annihilating filter and the DOAs are able to be obtained from the filter coefficients. To reconstruct the spatial annihilating filter, a multiple measurement vectors structured total least norm (MMV-STLN) approach is used under the structured total least squares (STLS) framework, along with a spatial fast Fourier transform (FFT) based initialization method. It is proved that the new method is equivalent to the deterministic maximum likelihood (DML) estimator of DOA. It can resolve up to M – 1 sources using M-element ULA and is capable of both incoherent and coherent signals. Numerical results show that the proposed method surpasses the representative methods largely in resolution and estimation performance under the low signal-to-noise ratio (SNR), snapshot deficient and closely spaced signals scenarios.

The non-collocated dual-polarized array alleviates the mutual coupling between the orthogonally polarized antennas. Nevertheless, the enlarged inter-element spacing may bring ambiguity issue, and the conventional direction-of-arrival (DOA) estimation algorithms may require varied processing approaches for different signal environments. For these problems, this paper first introduces two types of non-collocated array configurations called the interleaved array and the side-by-side array. The former helps to improve the aperture efficiency, and the latter can enable distributed applications. Subsequently, a polarization-agnostic DOA estimation method is proposed. In this method, firstly the low-dimensional array interpolation is performed via the joint minimization of the covariance matrix rank and a whitened residual error term. Then, both the auto- and cross-covariance matrices of the interpolated subarrays are employed to extract the signal subspace unrelated to the polarization parameters, based on which the fast direction finding algorithm is exploited. An alternating direction method of multipliers (ADMM) based solver is proposed to accelerate the array interpolation. The proposed method can handle the type of completely polarized (CP) signals, the type of partially polarized (PP) signals, the type of a spatial mixture of CP and PP signals, and the type of a temporal mixture of CP and PP signals without additional polarization-dependent processings. It is also compatible with the classical sparse array like the coprime array. Additionally, the extension to the 2D case is introduced. Numerical simulations are provided to demonstrate its superior performance in comparison to other representative approaches.

This letter considers the problem of underdetermined estimation of direction-of-arrival (DOA) and polarization parameters, including the degree of polarization (DOP), polarization orientation angle (POA), and polarization ellipticity angle (PEA), for partially polarized (PP) signals with a dual-polarized coprime array. Firstly, the increased dimension due to dual-polarized sensors is reduced by adding two covariance matrices, which facilitates the coarray construction. Then, the holes in the coarray of coprime array are interpolated via utilizing the annihilating relation. At last, the DOAs and array noise variance are obtained by solving an annihilating relation based optimization problem and then used to reconstruct the coherency matrices via the weight least squares (WLS). The polarization parameters can be estimated by performing eigendecomposition on the reconstructed coherency matrices. We show the underdetermined estimation of array noise variance which is difficult for conventional methods is vital for accurate underdetermined estimation of the polarization parameters. The performances of the proposed method are verified by numerical simulations and compared with other representative methods including the Cramer-Rao bound (CRB).

In this paper, a new wideband DOA estimation method for arbitrary array application is proposed. The arbitrary array manifold is approximately decomposed under the concept of manifold separation technique (MST) and a linear combination of the equal interval sampling complex exponentials (EISCE) can be synthesized. Since the EISCEs at all frequencies contain the same DOA information, they can be annihilated by the same spatial annihilating filter. So, the coherent annihilating scheme is proposed, and all narrowband components are coherently combined to construct an optimization problem under the structural total least squares framework. During the optimization problem construction, the general solution method is proposed to enable the model approximation error in MST to be reduced to a negligible level. Finally, the optimization problem is solved through the multiple measurement vectors structural total least norm (MMV-STLN) approach and the DOAs are estimated from the reconstructed spatial annihilating filter. The new method is free of frequency focusing and is capable of handling both incoherent and coherent signals. The simulation results verify that the proposed method surpasses the existing methods largely in resolution and estimation performance under the low SNR, snapshot deficient, and closely spaced sources scenarios.