Research On Sparse Channel Estimation For Massive MIMO Systems

Massive multiple-input multiple-output(MIMO)is a promising physical layer technology for 5G wireless communication.The application of massive MIMO needs to obtain accurate channel state information.With the increase of the number of antennas and the number of users in massive MIMO systems,traditional linear channel estimation methods such as LS,MMSE,etc.,require overwhelming training overhead and high computational complexity.In order to solve these problems,the most direct way is to utilize the inherent sparsity of the massive MIMO channel and reduce the effective dimension of the channel matrix.Based on the matrix completion and tensor completion theory,this paper studies low-overhead and low-complexity channel estimation technology in massive MIMO systems.The specific contents are as follows:Firstly,Channel estimation approaches applied for MIMO systems involve overwhelming training overhead and poor performance.In this paper,the Rice fading channel model is considered,and a new approach based on matrix completion is proposed.The channel estimation problem is formulated as a low-rank matrix completion problem by utilizing the inherent sparsity of the massive MIMO channel.The weighted nuclear norm is used to approximate the rank function,and then with the majorization-minimization(MM)algorithm to recovery the channel state information.Simulations demonstrate that,compared with the traditional algorithm and the existing low-rank channel estimation algorithms,the proposed algorithm can obtain better channel estimation performance and lower computational complexity with the same number of pilots.Secondly,many current channel estimation technologies are based on two-dimensional channel models,and the estimation efficiency and estimation accuracy need to be improved.This paper studies the channel estimation problem from the perspective of high-order data processing based on tensor completion,and establishes a three-dimensional tensor-type massive MIMO system channel model to complete the matching of the two-dimensional pilot matrix and the three-dimensional tensor-type channel state information.In addition,in order to solve the problem of huge channel estimation pilot training and feedback overhead in a multi-user massive MIMO system in FDD mode,a joint channel estimation scheme is adopted,that is,each user directly feeds the pilot signal sent by the base station to the base station,and the base station jointly estimates the channel state information.This scheme does not require the user to estimate the channel separately and then feed it back to the base station.Finally,in order to further improve the channel estimation efficiency and estimation accuracy of massive MIMO systems,this paper extends the algorithm and framework based on low-rank matrix completion to higher-order low-rank tensor completion problem.Based on the three-dimensional tensor channel model,this paper proposes a joint channel estimation method based on tensor completion.The channel estimation problem is formulated as a low-rank tensor completion problem by utilizing the inherent sparsity of the massive MIMO channel.The weighted kernel norm of the tensor is used to approximate the rank function,and then with the majorization-minimization(MM)algorithm to recovery the channel state information.Simulation results show that compared with existing 3D channel estimation algorithms,the algorithm has better estimation performance.