High-Resolution Millimeter Wave Channel Parameters Estimation Algorithm Based On SAGE And VB-SAGE

With the commercialization of the fifth-generation mobile communication,people’s requirements for the transmission rate and data flow of communication networks are increasing day by day.The frequency band below 6 GHz is becoming increasingly crowded,and existing spectrum resources are gradually unable to meet the growing communication needs.In this context,millimeter wave,with its large bandwidth and abundant spectrum resources,has come into people’s view.Therefore,a lot of work on millimeter wave channel detection and channel modeling has been carried out.However,due to the high frequency of millimeter wave and the extremely high path loss during propagation,massive antenna array and beamforming technology are often used to obtain high antenna gains for overcoming the large path loss and establishing reliable communication links.In order to achieve multi-stream and multi-user transmission close to digital beamforming with relatively low hardware overhead,hybrid beamforming architecture is usually adopted in millimeter wave massive Multiple Input Multiple Output(MIMO)systems.In the hybrid beamforming architecture,the number of RF links is much smaller than the number of antenna elements,which compresses the data dimension,changes the antenna data structure,and poses great challenges for channel estimation.This article focuses on the application scenario of massive MIMO millimeter wave channel detection and conducts research on channel multipath parameters estimation.The main contents and innovations are as follows:1.The propagation characteristics of millimeter wave channels,such as high path loss,sparsity in delay and angle domains are introduced.The adopted geometry-based large-scale MIMO parameterized channel model for millimeter wave channels is described.The commonly used direction of arrival estimation algorithms in channel parameters estimation,as well as the Space-Alternating Generalized Expectation-Maximization(SAGE)algorithm and Variable Bayesian(VB)-SAGE algorithm frameworks involved in this article are summarized,laying a theoretical foundation for subsequent research.2.Considering that the system adopts broadband massive MIMO antenna configuration,beamforming technology on transmitter and receiver,and frequency domain excitation for channel detection,a two-stage high-precision channel parameter estimation algorithm based on SAGE is proposed,starting from a geometric parameterized channel model.This algorithm uses serial interference cancellation in the initialization stage to achieve path-by-path parameters estimation,and parallel interference cancellation in the iterative update stage to achieve path parameter refinement,which solves the angle mismatch problem in the original SAGE algorithm and significantly improves estimation performance.The algorithm estimates the angles by coarse search and refinement,and does not depend on the specific antenna array form.The simulation verifies that this algorithm can achieve high estimation accuracy when the receiving and transmitting ends are equipped with uniform linear arrays and uniform rectangular arrays.3.A joint estimation algorithm based on VB-SAGE for channel path number and path parameters is proposed to address the issue that the path number is assumed to be known in most channel estimation studies.Starting from the signal model and probability model,variational expressions are derived within the framework of the VB-SAGE algorithm.Two path selection schemes based on Gaussian and Laplacian priors,as well as two estimation algorithms,VB-SAGE-G and VB-SAGE-L,are proposed.The computational complexities of two algorithms are analyzed,and the good performance of the algorithm in terms of path number estimation,path parameters estimation,and channel matrix estimation was verified through simulation.In addition,experiments have shown that the use of Laplacian priors to remove pseudo paths in noise is more effective and can improve the estimation performance.