Characteristics Analysis And Modeling Of Multi-frequency Wireless MIMO Channels

2019 witnessed the first step in commercialization of Fifth-Generation（5G）global communication system.Following the tradition of“planning the next generation of mobile communication system while commercializing the current one”,the research on Sixth-Generation（6G）has been rolled out.The vision of 6G:enabling seamless global coverage to ensure safe and reliable connections among people,machines and things in any place at any time,will guarantee exponentially increasing data amount.With frequency resources getting rarer,multi-frequency collaboration will become an inevitable trend,therefore studying multi-frequency channels is of great importance.To meet this demand,this thesis focuses on conducting the multi-frequency channel measurement,proposing channel similarity indices and investigating similarities between measured multi-frequency channels,and evaluating current multi-frequency channel models.This thesis contains the following innovative work:（1）An outdoor marco-cell channel measurement is conducted using two sets of channel sounders simultaneously on three frequency bands:0.7 GHz,2.3 GHz,and 3.7 GHz（0.7 GHz and 3.7 GHz are currently used in 5G）.Perfectly identical environments for two frequency bands are created through synchronized trigger signals between the two channel sounders and through placement of two antenna arrays.The principles of channel calibration,description of channel sounders and detailed planning of the measurement campaign are given,which are the premise of obtaining correct data.（2）Measurement data are validated from two aspects.Then,multi-path components are extracted by using Space-Alternating Generalized Expectation-Maximization（SAGE）algorithm,which,later on,are clustered by algorithms based on KPower Means.Channel parameters are calculated following the processing of data.By referring to non-stationary channel modeling,similarity indices for multi-frequency channels are proposed covering three dimensions:time,space and frequency.An analysis on measured channels reveals that the similarity of multi-freuqency channels are high both in Line-of-Sight（Lo S）and Non-Line-of-Sight（NLo S）scenario,in terms of the frequency domain,while it is only high in Lo S scenario in terms of space domain.Besides, after comparing the channels with different frequency gaps,the phenomenon that the smaller the frequency gap between two channels,the more similarity they share has not been obsevered.（3）A thorough description is given on the multi-frequency modeling approach of two standardized channel models:3^rd Generation Partnership Project（3GPP）TR 38.901and QUAsi Deterministic Rad Io Channel Gener Ator（Qua DRi Ga）,which focuses on the generation approach of small- scale parameters.After inputting the parameters extracted from measurement to evaluate the two models.In addition,two modifications are proposed to 3GPP TR 38.901 channel model to make it more flexible.To better model the time-nonstationarity,3D space-time-frequency non-stationary wideband MIMO channel is introduced along with two mechanisms to characterize time non-stationary channels.And its superiority is demonstrated compared to the two channel models mentioned above.This model can be applied to model time non-stationary multi-frequency channels.At last,multi-frequency channels are used as Channel State Information（CSI）in a robust precoding algorithm.it is found that the sum rate increased compared to that by precoding schemes that use independently generated channels.