Traditional beam tracking has severe performance loss under the high mobility and narrow beam scenario. To alleviate the tracking performance degradation, we propose an adaptive beamwidth control for millimeter wave beam tracking. Using the particle filter, we perform the beam tracking, i.e., which is to estimate the channel gain and the angle-of-arrival (AoA). Based on the updated estimate standard deviation of AoA, the adequate beamwidth is set via the partial activation of the antenna array. Simulation results show that the beam tracking with the proposed beamwidth control maintains a higher received signal-to-noise ratio (SNR) than that of the beam tracking without the beamwidth control while the SNR gain leads to a smaller AoA tracking error under high mobility.

In this paper, we propose location-aware beam training and multi-dimensional atomic norm minimization (ANM)-based channel estimation for reconfigurable intelligent surface (RIS)-aided millimeter-wave systems. The use of both location information and RIS beamwidth adaptation allows a significant reduction of beam training overhead. However, considering a trade-off between accuracy and beam training overhead, this may induce inaccurate channel estimation. Nevertheless, superior channel estimation performance is achieved by multi-dimensional ANM techniques, which have been shown to be effective in capturing cascaded structures such as the channel in RIS-aided systems. In the proposed work, a cascade of BS-to-RIS channel and RIS-to-BS channel is represented as a linear combination of either steering vectors, 2D steering vectors, or 3D steering vectors, and ANM with appropriate dimension is applied to estimate the channel. From simulation results, it has been demonstrated that location-aware channel estimation via 2D ANM and 3D ANM achieves excellent estimation accuracy along with a reduced beam training overhead.

Reconfigurable intelligent surface (RIS) is an emerging technique for robust millimeter-wave (mmWave) multiple-input multiple-output (MIMO) systems. In this paper, we study the channel estimation problem for extremely large-scale RIS (XL-RIS) assisted multi-user XL-MIMO systems with hybrid beamforming structures. In this system, we propose an unified channel estimation method that yields a notable estimation accuracy in the near-field BS-RIS and near-field RIS-User channels (in short, near-near field channels), far-near field channels, and far-far field channels. Our key idea is that the effective channels to be estimated can be each factorized as the product of low-rank matrices (i.e., the product of a common matrix and a user-specific coefficient matrix). The common matrix whose columns are the basis of the column space of the BS-RIS channel is efficiently estimated via a collaborative low-rank approximation (CLRA). Leveraging the hybrid beamforming structures, we develop an efficient iterative algorithm that jointly optimizes the user-specific coefficient matrices. Via experiments and complexity analysis, we verify the effectiveness of the proposed channel estimation method (named CLRA-JO) for the three categories of wireless channels.

This paper presents an efficient channel estimation algorithm for multi-user reconfigurable intelligent surface (RIS)-aided millimeter-wave (mmWave) systems. In this paper, the concept of low rank matrix completion (LRMC) is exploited to reduce beam training overhead for channel estimation. The proposed beam training samples part of each channel matrix in a special pattern that is suitable for LRMC with less beam training overhead. Then, the beam training is followed by multi-user channel estimation. For computationally efficient channel estimation, the proposed algorithm exploits the property that all the channel matrices share the same low-rank subspace in multi-user RIS-aided systems. The shared subspace is derived by combining candidate subspaces, which are estimated by fast alternating least squares (FALS) from partially observed channels. With the shared subspace, all the missing entries of channels are recovered via computationally efficient linear estimation. The simulations and complexity analysis demonstrate that the proposed algorithm shows a superior accuracy-complexity trade-off compared to existing works.

This paper introduces an off-grid DoA estimation via two-stage cascaded network which can resolve a mismatch between true direction-of-arrival (DoA) and discrete angular grid. In the first-stage network, the initial DoAs are estimated with a convolutional neural network (CNN), where initial DoAs are mapped on the discrete angular grid. To deal with the mismatch between initially estimated DoAs and true DoAs, the second-stage network estimates a tuning vector which represents the difference between true DoAs and nearest discrete angles. By using tuning vector, the final DoAs are estimated by moving initially estimated DoAs as much as the difference between true DoAs and nearest discrete angles. The limitation on estimation accuracy induced by the discrete angular grid can be resolved with the proposed two-stage network so that the estimation accuracy can be further enhanced. Simulation results show that adding the second-stage network after the first-stage network helps improve the estimation accuracy by resolving mismatch induced by the discretized grid. In the aspect of the implementation of machine learning, results also show that using CNN and using PReLU as the activation function is the best option for accurate estimation.

This paper presents an experimental validation of deep learning-based direction-of-arrival (DoA) estimation by using realistic data collected via universal software radio peripheral (USRP). Deep neural network (DNN) and convolutional neural network (CNN) structures are designed to estimate the DoA. Two types of data are used for training networks. One is the data synthesized by the signal model, and the other is the data collected by USRP. Here, the signal model considers both mutual coupling and multipath signals. Experimental results show that the estimation performance is most accurate when training DNN and CNN with the collected data. Furthermore, the estimation tends to be poor in the indoor environment, which suffers from the strong non-line-of-sight (NLoS) signals.