Research On Statistical CSI Acquisition In Nonstationary Massive MIMO Environments

As an extension of the traditional multiple-input multiple-output(MIMO),massive MIMO with hundreds or thousands of antennas at the base station(BS)can effectively increase system capacity.To benefit from advantages of massive MIMO,it is necessary to obtain an accurate statistical channel state information(S-CSI).Due to the large size of the antenna array and the birth-death process of scatterers,massive MIMO channel has a particular nonstationary characteristic,which is different from the conventional MIMO channel.However,the nonstationarity leads to the S-CSI to change with time,so that the traditional S-CSI acquisition methods based on the wide-sense stationary assumption are no longer suitable.Therefore,this dissertation is concerned with the S-CSI acquisition in nonstationary massive MIMO environments.Firstly,the dissertation studies the S-CSI acquisition in the temporal nonstationary massive MIMO environment,and a novel hidden statistical channel state Markov model(HSCSM-model)is established.The parameter of the HSCSM-model is estimated through learning the observed sequence of received signals.Based on the proposed model and its estimated parameter,the S-CSI can be obtained through a maximum a-posteriori decision process(MAPDP).Simulation results show that an accurate S-CSI acquisition can be achieved by the proposed method in the nonstationary massive MIMO environment.In addition,the estimation accuracy rate of the proposed method increases with the length of observation sequence as well as the number of antennas,where a tradeoff between them exists given a limited computing ability/storage space.Then,the research on S-CSI acquisition is extended to the spatio-temporal double nonstationary massive MIMO environment,and a new statistically double nonstationary channel model(SDNCM)is proposed.Based on the proposed model,a parallel estimation method(PEM)and a successive estimation method(SEM)are proposed.Moreover,a user power allocation policy is introduced to improve the estimation accuracy rate of the proposed methods.For the S-CSI acquisition problem,simulation experiments compare the performance of two methods,and the computational complexities of two methods are analysed.Simulation results show that the computational complexity of the SEM is significantly improved compared with that of the PEM.Finally,the main contributions are summarized and future works are described.In summary,the research topic considers the S-CSI acquisition problems in the temporal and spatio-temporal double nonstationary massive MIMO environments,respectively,corresponding statistical models and S-CSI acquisition methods are proposed,obtaining an accurate S-CSI is beneficial to improve the performance of massive MIMO communication systems.