Research On Channel Information Acquisition Technology Of Massive MIMO System

Recently,the popularity of smart mobile terminals rapidly increases the data demand so the network need improve the capacity significantly.Massive multiple input multiple output(MIMO)technology has been widely studied and proved to realize the multi-user multiplexing in the space domain,thus it has become the most important technology for the capacity improvement.In wireless communication system,the channel state information is the base of signal detection.Because the massive MIMO system is equipped with many antennas,so the channel state information(CSI)acquisition is much more important compared with MIMO system.The thesis focus on the CSI acquisition technologies in Massive MIMO system.Firstly,the massive MIMO channel model,pilot structure and a massive MIMO system example are introduced.Based on the Rayleigh channel model and orthogonal pilot structure,the channel estimation methods and uplink,downlink data transfer procedure are presented.The system achievable rate under limited pilot resource are analyzed.The results of the analysis show that the capacity of uplink and downlink are mainly influenced by the so-called pilot contamination.Then,the thesis studies two methods about the improvement of the semi-blind pilot design.By changing the structure of semi-blind pilot design,a method which combine the user grouping and semi-blind pilot design is studied.The channel estimation,uplink and downlink capacity are derived.The simulation about the performance of semi-blind pilot design and orthogonal pilot design are conducted.The results tell us that the semi-pilot design combined with user grouping can improve the capacity,decrease the propagation of estimation error and has capacity gain in a longer coherent time compared with orthogonal pilot design.From the perspective of capacity optimization,a method about the power allocation of pilot and data is studied.The simulation results show that at low signal to noise ratio(SNR),the power allocation method can improve the system capacity.Subsequently,the thesis introduces a data-aided channel estimation scheme focused on the problem about channel estimation error propagation of semi-blind pilot design.The scheme contains two typical methods which are hard detection based estimation and soft detection based estimation.The improvement and con-vergence about soft detection aided channel estimation are analyzed and the simulation results present the relationship between variance of soft decision symbol and SINR of detection.Besides,from different perspec-tives,the simulations of the data-aided iterative channel estimation and pilot only based channel estimation are conducted.The numerical results show while the data is accurate and the length of data is long then the data-aided channel estimation has better performance.Besides,after several iteration the channel estimation is convergent,the normalized mean square error(NMSE)of different users tend to be the same.Finally,considering the sparseness of channel state information in massive MIMO angle domain(or beam domain),a discrete Fourier transform aided channel estimation method are studied.DFT-aided channel estimation transfer the LS estimation results to angle domain by DFT operation and filter the result in angle domain.The simulation results show that with the information of DOA,the performance of DFT-aided es-timation is better than LS estimation and is near to MMSE estimation.However,the DFT-aided estimation has estimation error floor at high SNR.Based on the DFT-aided estimation,the thesis introduce a pilot reuse algorithm.The pilot reuse algorithm can significantly reduce the pilot resource consumption.At low and medium SNR,the performance of DFT-aided channel estimation with pilot reuse is near to MMSE estimation with orthogonal pilot.