Research On Massive MIMO Channel Models And Its Non-stationary And Joint-correlation Characteristics

With the commercial application of fifth generation communication(5G),massive MIMO system,in which a base station equipped with hundreds of antenna elements,has emerged as a promising technique of 5G and has drawn considerable attention.Massive MIMO technique can potentially make full use of spatial freedom to reap a significant increment of transmission rate and spectral efficiency.In order to evaluate and study the performance of massive MIMO systems,it is essential to investigate the corresponding channel models.In this thesis,near-field effect and non-stationary characteristics of massive MIMO channel are analyzed.Besides,a novel massive MIMO channel model based on extended joint-correlation model is proposed and investigated.Firstly,the near-field effect of massive MIMO channel is investigated.Due to the large number of antenna elements,near-field effect is prone to occur in massive MIMO systems and plane wave hypothesis is no longer satisfied under near-field condition.From the perspective of the accuracy of channel modeling,spherical wave is supposed to be adopted instead of plane wave to describe the propagation.In this thesis,the properties of plane wave and spherical wave schemes are compared.On this basis,the application ranges of plane wave and spherical wave schemes are proposed by analyzing eigenvalue distribution of channel matrix.The simulation results show that the inaccuracy caused by plane wave hypothesis cannot be ignored in massive MIMO systems.Spherical wave can get the utmost out of performance gain brought by massive MIMO technique.Subsequently,the non-stationary characteristics of massive MIMO channel are researched.In massive MIMO systems,the scattering environments experienced by different antenna elements are variant in space,which reflects the spatial non-stationary characteristics.In addition,the time-varying channel environment leads to the non-stationary on time axis.In this thesis,birth-death processes on time axis are extended to space domain to investigate non-stationary characteristics on array axis based on the twin-cluster geometric-based channel model.The entire procedure of modeling 3D space-time non-stationary characteristics is proposed.The numerical results of received power and correlation show that this procedure of birth-death processes in twin-cluster channel model can capture the non-stationary characteristics of 3D massive MIMO channels.Finally,the application of joint correlation model in massive MIMO channel modeling is investigated.Conventional Kronecker model assumes that transmitter and receiver are independent,while joint-correlation model assumes that they are related and a spatial coupling matrix is used to illustrate this correlation between transmitter and receiver.In this thesis,the influence on channel performance of different power azimuth spectrum is investigated.Besides,we proposed an extended joint-correlation channel model focused on both non-stationary and joint-correlation characteristics in massive MIMO channels.Furthermore,the performances of extended joint-correlation channel model proposed in this thesis and Kronecker channel model are compared in massive MIMO scenario based on analysises of condition number,correlation matrix distance(CMD)and channel capacity.The numerical results show that extended jointcorrelation model proposed in this thesis can describe the characteristics of massive MIMO channel more accurately than Kronecker model.