Low Complexity Channel Estimation Algorithm Under Massive MIMO

In recent years,massive MIMO technology,as a key technology to provide services for various 5G application scenarios,has become the focus of the industry.The accuracy and complexity of channel estimation are the key factors to determine whether massive MIMO technology can fully realize its potential.With the increase of the number of antennas and users,the dimension of channel matrix to be estimated increases continuously,and the complexity of channel estimation increases with the dimension cubic times of channel matrix,so channel estimation is faced with severe challenges.Therefore,this dissertation studies the uplink channel estimation problem in massive MIMO multi-user systems and proposes a low-complexity channel estimation method.The main work includes the following two points:(1)Based on the joint sparse characteristics of multi-user channels in angle domain and time delay domain,a structured joint channel estimation(UG-SJSCE)algorithm based on user grouping is proposed.It adopted a grouping method based on user angle domain support agglomeration class,the purpose is to realize the user angle domain support set of precise positioning,through recovery angle domain only support column vector to locate the user on the set time delay support set(user sparse path)effectively,realize to estimate the channel dimension reduction so as to reduce the computational complexity.Using MATLAB simulation tool,the effectiveness of UG-SJSCE algorithm and its performance superiority over traditional compressed sensing algorithm are verified from different angles.(2)Aiming at the unpredictability of user channel sparsivity and the sensitivity of estimation performance to the accuracy of user grouping in the UG-SJSCE algorithm,a joint user grouping and channel estimation algorithm based on the variable db esian framework(UG-VBL)was proposed.In this algorithm,the channel estimation problem is summarized as a statistical inference problem.The channel impulse vectors and group vectors are regarded as hidden variables to be estimated.The variational method is used to iterate the approximate posterior probability distribution of the optimized hidden variables.Furthermore,a UTAMP algorithm based on unitary transformation is adopted to avoid matrix inversion and simplify the computation for the matrix inversion with high complexity in the iterative steps of the algorithm.By drawing the simulation curve and comparing the running time of the algorithm,the MATLAB simulation tool is also used to verify that the UG-VBL algorithm can achieve lower estimation error and lower computational complexity.