Research On Several Key Technologies In Massive MIMO Systems

Recently,the massive multiple-input multiple-output(MIMO)has become a key technology for next-generation wireless communication systems because of its large spec-tral efficiency and energy efficiency.To realize the advantages of massive MIMO,the acquisition of channel state information is important.On the other hand,due to the hard-ware imperfection,the effect of phase noise in massive MIMO systems is not negligible.Therefore,this thesis focuses on the channel estimation and phase noise suppression in massive MIMO systems.For the channel estimation in massive MIMO,the non-stationarity of massive MIMO channels is considered,which means that the sparsity pattern of the channels varies along the base station antenna array.Due to the limited number of scatterers in the physical en-vironment,the massive MIMO channels are typically sparse in the delay domain.Mean-while,since the distance between adjacent antennas at the base station is small,different channel vectors tend to have a similar sparsity pattern.In this thesis,a Dirichlet Process(DP)and Gaussian hybrid prior model is designed for the channel vectors to characterize both the common sparsity and the individual sparsity of the channels.Under the varia-tional Bayesian inference framework,the channels can be adaptively divided into several groups,each group bearing a similar sparsity pattern(i.e.,the same common sparsity pat-tern).The common sparsity can be utilized to improve the channel estimation performance significantly.For the phase noise suppression in massive MIMO,the statistical property of the phase noise is used by applying the eigenvalue decomposition for the covariance matrix of the phase noise vectors.According to some approximations,the number of unknown parameters to be estimated can be significantly reduced.In this phase noise suppression problem,a two-stage protocol is considered in the communication process,i.e.,the chan-nel estimation stage and the data transmission stage.In the channel estimation stage,a Gaussian prior is assigned to the channel vectors to control the sparsity pattern.Under the variational Bayesian inference framework,the posterior probabilities of the channel vectors and the phase noise sequences can be iteratively computed,and the posterior mean of the channel vectors is used as the estimate.Similarly,in the data transmission stage,the posterior probabilities of the data symbols and phase noise sequences are computed,and the posterior mean of the data symbols is used as the estimate and for data demodulation.To verify the effectiveness of the proposed channel estimation and phase noise sup-pression algorithms,extensive simulations are conducted evaluating the mean square er-ror and the bit error rate of the system.The simulation results show that the proposed algorithms can achieve good performance and significantly outperform the other existing algorithms.