| The rapidly emerging mobile social network and mobile multimedia services have put forward more stringent requirements on the improvement of data transmission rate and the realization of diversified application types.Therefore,in order to meet the increasingly stringent mobile communication needs of users,massive multiple-input multiple-output(MIMO)technology has emerged.Frequency division duplex(FDD)is one of the most widely used Duplex modes in mobile communication systems.However,the use of largescale MIMO brings challenges to the estimation of downlink channel state information(CSI)in FDD systems.Due to the lack of information reciprocity between uplink and downlink in FDD system,downlink CSI is usually obtained through downlink pilot training and feedback.However,the base station(BS)of large-scale MIMO cannot design completely orthogonal pilot for a large number of antenna ports.Therefore,new ideas are needed to solve this problem.According to the above problems,this paper will study the theory and key technologies of massive MIMO wireless transmission system from the aspect of complexity of the channel estimation method,the pilot overhead of downlink reconstruction,the channel characteristics of dual-polarized system and multi-user system,and looking for solutions to existing large-scale MIMO technology problems at the same time.Firstly,we study the principle of MIMO system and investigate the existing channel reconstruction methods.The basic model of wireless channel,the concepts of MIMO system and the principles of Orthogonal Frequency Division Multiplexing(OFDM)technology are introduced in detail,and the existing channel reconstruction and channel extrapolation methods under FDD mode are investigated and summarized.The problems such as high complexity and high pilot cost of downlink channel reconstruction in FDD mode are proposed,which lays a foundation for the design of channel reconstruction method for dual-polarized system and multi-user MIMO system.Secondly,we study the channel reconstruction method based on polarization independent characteristics for dual-polarized MIMO systems and propose a dual-polarized channel extrapolation scheme based on the path number,delays and angles of arrival of the propagation paths.According to the spatial similarity between channels in different polarization directions,the user emits a large number of pilots in the initial polarization direction,and the polarization independent parameters of the channel are estimated by using The newtonized orthogonal matching pursuit(NOMP)algorithm in BS.Then,the polarization independent parameters of the extrapolated channel are refined with a small number of pilot,and reconstruct the channels in other polarization directions.Furthermore,considering the non-perfect reciprocity of channels in different polarization directions,we design an extrapolation compensation method based on the energy of each path.The simulation results show that the dual-polarized channel extrapolation method is effective and cost-effective,and the experimental data verify that the scheme can also be well applied in the NLoS propagations.Then,we study channel reconstruction methods based on spatial similarity characteristics of adjacent user channels for large-scale MIMO systems in multi-user scenarios,and design a user uplink channel extrapolation scheme,which includes user scheduling module and user extrapolation module.In the first user scheduling module,the users in the cell are grouped by comparing the similarity of the direction diagram of the user uplink channel,and the neighboring users are clustered into a group,which is highly flexibility;the independent user extrapolation of each group is carried out according to the result of user scheduling module in the following user extrapolation module.In the user extrapolation module,NOMP algorithm is used to accurately estimate the initial user channel in the group,and then a smaller size dictionary is derived from the initial user estimation result.Then CSI of the remaining users in the group is seeked inside the smaller size dictionary with half of the pilot and the uplink channel of users is reconstructed according to these extrapolated CSI parameters.The simulation results show that the user extrapolation scheme can effectively reconstruct multi-user channels with low computational complexity and pilot cost,and has good performance in the scenario of large number of users in the cell wuth large distance between users.Finally,we study a downlink reconstruction method with low overhead is studied for FDD multi-antenna system,and propose a downlink CSI optimization and quantitative feedback method based on spatial reciprocity between uplink and downlink in FDD mode.NOMP algorithm is used to estimate the frequencyindependent parameters of the uplink,including the delay and angle of arrivals of each path,and they are quantized to the lattice points of the oversampled code book in NOMP algorithm as the initial parameters.After that,only a few receiving antennas are used to transmit the downlink pilots and the estimated uplink CSI,and the initial parameters are optimized according to the downlink receiving signal at the user equipment side.We further design a high-precision quantization method based on oversampling codebook.The delay,angle of arrival and gain of each path are quantized at the user side,and the results are feeded back to BS,where the channel reconstruction of downlink is completed.The simulation results show that the low-overhead downlink reconstruction method based on quantization method has excellent channel reconstruction performance,and it can fit communication scenarios with CSI fluctuation between uplink and downlink well. |
