Research On Channel State Information Acquisition And Feedback Technology In FDD Massive MIMO Systems

Massive MIMO technology which has become one of the key technologies of the fifth generation mobile communication system(5G)can improve spectrum efficiency and enhance system capacity greatly.Theoretically,in the downlink channel of Massive MIMO systems,based on the knowledge of channel state information(CSI),the precoding technology can eliminate the inter-user interference and improve the system performance effectively.However,the acquisition of CSI will take up a lot of time and frequency resource for frequency division duplex(FDD)Massive MIMO systems.This thesis focuses on the CSI acquisition and feedback technology in FDD Massive MIMO systems.Detailedly,we research the algorithm of noncoherent trellis-coded quantization(NTCQ),the scheme of joint space division multiplexing(JSDM)and the technology of joint orthogonal matching pursuit(J-OMP).Firstly,precoding techniques of Massive MIMO systems are summarized.The CSI acquisition methods are briefly discussed,and the problems of CSIT acquisition in FDD systems are pointed out.Simulations and analysis of precoding technique based on the ideal CSIT and the limited feedback technique based on the non-ideal CSIT in Massive MIMO systems are presented respectively.Secondly,the NTCQ algorithm under different channel conditions are realized.The system model and quantization process of NTCQ algorithm in Rayleigh fading channel are introduced and analyzed.The Diff-NTCQ algorithm is implemented in the condition of the time correlated channel.The Space-NTCQ improved algorithm is proposed for the spatially correlated channel.Simulation results show that the NTCQ?Diff-NTCQ and Space-NTCQ algorithms can effectively reduce the time complexity of the quantization process,and has good flexibility and scalability for different number of antennas.Thirdly,We analyze the processing of JSDM transmission schemes with different user grouping scenarios.The system model and design process of the JSDM scheme under single ring model are presented,when the users are grouped ideally.Aiming at the non-ideal user grouping scenario,a new JSDM improved algorithm based on K-Medoids user grouping method is proposed.The performance and parameter optimization of JSDM and improved JSDM algorithm are simulated and analyzed.Simulation results show that the PGP-JSDM transmission method can achieve significant reduction of CSI feedback.The improved JSDM algorithm based on K-Medoids user grouping method can realize user grouping adaptively.Finally,the estimation and feedback algorithm of CSIT based on compressed sensing technique is studied.The joint sparsity properties of the virtual angular domain channel in the single user and multi-user system are discussed respectively.Based on the common sparse characteristics among multiple fading blocks,a common sparsity of multiple fading blocks based J-OMP(CSMFB-J-OMP)algorithm is proposed.The J-OMP and CSMFB-J-OMP algorithms are simulated and analyzed.Simulation results show that the J-OMP algorithm has good NMSE performance,and the improved CSMFB-J-OMP algorithm can effectively reduce the overhead of pilot training and uplink channel feedback while maintaining good NMSE performances.