Research On Channel Estimation Based On Mathematical Statistics Method In Massive MIMO System

Massive Multiple-Input,Multiple-Output(MIMO),as one of the key technologies of the next generation mobile communication,has been extensively studied.The prominent feature of the massive MIMO system is that Base Station(BS),equipped with a large number of antennas,can serve several single-antenna users simultaneously at the same frequency band.Research results have shown that it can offer an increased spectral efficiency through spatial multiplexing.In addition,with simple linear signal processing,such as Zero-Forcing(ZF)or Minimum Mean Square Error(MMSE),it can not only improve the data transmission rate,but also improve the energy efficiency.The above-mentioned advantages of the massive MIMO system are based on a fact that the BS can accurately estimate the Channel State Information(CSI)of the system.In general,CSI is unknown to the system.A conventional way to acquire the CSI is training-based channel estimation,that is,each user transmits a pre-designed,orthogonal pilot sequence,and then,the base station uses the received signals to estimate CSI.But for the massive MIMO system,the number of users is very large.To ensure the orthogonality of the pilot sequences,they must be very long and it consumes a lot of time-frequency resources.Moreover,in a multi-cell,massive MIMO system,co-channel cells will be set up at a close range.Hence,when using time-multiplexed pilots,the requirement for high spectral efficiency necessitates sharing of pilot sequences between adjacent cells,resulting in the channel estimates of the users in a cell being corrupted by the channel vectors of users in the adjacent cells.Thereby the accuracy of the CSI estimation will be degraded drastically.Futuremore,this inter-cell interference does not tend to disappear with the number of antennas increase.This phenomenon is known as pilot contamination,and has become a bottleneck for the performance of massive MIMO systems.Recently,a channel estimation method based on Superimposed Pilot(SP)has been introduced to the massive MIMO system.In this method,no time slot are cost on training.Hence,transmission rate of the system is higher than the traditional pilot-based method.In this thesis,we propose a new channel estimator for the massive MIMO system,which is based on the SP channel estimation method.And then strictly follows the definition of the MMSE estimation,it turns out that a matrix is needed to multiply on the right hand side of the existing estimators.Further,a new decoder is designed based on the MMSE decoder.Finally,simulation results show that the performance of the system with proposed method is superior to the traditional training methods and existing SP methods.This thesis also proposes a channel estimation method based on SP and the second-order statistics for the massive MIMO system.In this proposed method,the BS uses the Principle Component Analysis(PCA)to reduce the dimension of the received signals,and uses the Singular Value Decomposition(SVD)to get an estimation of part of the channels.Then,by combing the previous SP-based estimation and the SVD,a new estimation is obtained.Finally,a decoder similar to the MMSE decoder is also designed.Simulation results show that the performance of the system with the proposed method is superior to other methods.