Research On Spatial Non-stationary Characteristics And Statistical Modeling Of Large-scale MIMO Channels

It is of great significance for channel model to evaluate the mobile communication system.Given that large-scale MIMO technology is currently recognized as the next generation communication technology with huge potential,it is necessary to study its channel characteristics and modeling.There are two characterastics making large-scale MIMO channels different from conventional MIMO channels,that is,the non-stationaries over the array and the near field effects,which make some of the assumptions of the conventional MIMO channel model no longer appropriate.Therefore,modeling the large-scale MIMO channels based on expanding conventional model theory has become a hot topic at present.Firstly,a two-dimensional confocal elliptical model is used to discuss in several aspects: how many aspects the non-stationary characteristics in channel model will represent with the increasing of the antenna elements.And compared with conventional MIMO channel model,how is the varation tendency of channel performance when assuming that the visible scatterers differences in various antenna elements from the same cluster are different.Numerical results have demonstrated that there are some differences in the angle of arrivals and Doppler frequencies of different elements on the array.According to the measurement,the spatial correlation coefficients have been analyzed.Simulation shows that the visible scatterers difference among antenna elements in the same cluster significantly reduces the spatial correlation of array elements.Secondly,we studied the correlation modeling of large-scale MIMO channels based on the Wechselberger model: considering that the two-dimensional confocal elliptical model covers the near-field effects and non-stationaries over the array,its channel coefficients can be taken as “quasi-measurements”.The coupling matrix parameters are extracted from the channel matrix,and by analyzing the statistical properties of the coupling matrix,an empirical formula for randomly generating the coupling coefficient is given to approximate the “quasi-measurements” channel matrix.And then according to the non-stationary characteristics of the channel,a geometry-based spatial correlation calculation method is derived.Finally,based on the empirical formula of coupling matrix and the spatial correlation,an empirical model of the random environment is proposed for large-scale MIMO channels.The spatial correlation,channel capacity,and condition number distribution of the empirical model are investigated separately during the research.It shows that the empirical model can describe the characteristics of large-scale MIMO channels correctly.