Statistical Channel State Information Acquisition For Massive Mimo Wireless Communication

With the rapid popularization of intelligent terminals and a large increase in the demand of mobile new services, massive multi-input multi-output (MIMO) technology is attracting more and more attention. Mas-sive MIMO system can meet the communication demands of multiple users by equipping large antenna array at base station. The spectral and power efficiency can be greatly improved by exploting spatial dimension resources. In this thesis, the statistical channel state information (SCSI) acquisition for massive MIMO wire-less communication is dealt with. Specifically, approximate message passing (AMP) based SCSI acquisition methods are investigated.Firstly, a brief exposition of the factor graph and the expectation maximization (EM) algorithm is given. The factor graph can describe a global function when the global function is decomposed into the product of several local functions. The methods based on the factor graph to calculate the single edge function and all edge functions are introduced. The AMP algorithm based on the factor graph is introduced. The whole process of estimating parameters of the probabilistic model with hidden variables by using the EM algorithm and the convergence is given. The EM algorithm is applied to estimate the parameters of the Gauss mixture (GM) model.Subsequently, the SCSI acquisition in massive MIMO system of narrow band is investigated. The beam domain channel model and the SCSI required in AMP channel estimation for massive MIMO system is p-resented. In the phase of SCSI acquisition, the channels between all users and the same beam at the base station are modeled by GM distribution and the AMP algorithm and the EM algorithm are combined to es-timate the channel parameters. The SCSI can be acquired by averaging the sample CSI. An improved SCSI acquisition method is proposed. It estimates the average angle of arrival (AOA) and angular spread (AS) after acquiring primitive SCSI and obtains the power azimuth spread (PAS) distribution. The improved SCSI is acquired through the PAS. Simulation results show that the performance of the AMP channel estimation using the estimated SCSI can approach that of using the ideal SCSI, and the improved method can achieve better performance compared with the conventional method.Finally, the SCSI acquisition in the massive MIMO OFDM system is investigated. For the massive MIMO OFDM system, the channel model and space characteristics are analyzed and the beam domain channel model is presented. The SCSI required the user scheduling for the beam division multiple access (BDMA) transmission scheme is given. The SCSI acquisition method applying the uplink sounding signal is proposed. The Zadoff-Chu (ZC) sequence is used to generate the sounding signal and the sounding signal resource mapping is designed. The SCSI acquisition method for massive MIMO OFDM system is proposed after using the AMP algorithm and the EM algorithm to estimate the multi-path channel parameters. Simulation results show that the proposed multi-path channel parameters estimation method can estimate multi-path channel parameters accurately, including path delay and power. The eigenmode channel coupling matrix and the noise variance can be estimated with relative accuracy using the proposed SCSI acquisition method, and the user scheduling using the estimated SCSI can achieve good scheduling performance.