Compressed Sensing Based Sparse Channel Estimation And Precoding For Massive MIMO System

The wide application of diversified mobile intelligent devices promotes the demand for high data transmission rate.Large bandwidth is a feasible scheme to achieve huge throughput.Due to the shortage of existing communication spectrum resources,people extended the research to the free millimeter-wave band with higher frequency.However,millimeter-wave carrier requires more antennas to overcome path loss.Fortunately,high frequency of carrier corresponds to small size antenna,which is conducive to the deployment of large-scale antenna array.A new complex hybrid structure is adopted in millimeter-wave large-scale MIMO communication system,which is different from that of traditional wireless communication system,due to the limitation of hardware.Under this new structure,accurate channel estimation and reduction of interference between different users have become two major research directions.Channel estimation can obtain accurate channel parameters,guiding both transmitters and receivers to process the signals according to the current channel state to ensure the accuracy of the received signals.Precoding technology can suppress the interference between multiple user data streams according to the channel state information.Hybrid precoding technology is used in millimeter-wave large-scale MIMO communication system,which is dedicated to approaching the good performance of traditional all-digital precoding scheme,ensuring the correct transmission of signals.The main work and innovations of this thesis are described as follows.In the aspect of channel estimation,a millimeter-wave massive MIMO communication system model based on lens antenna is used to transform the traditional channel into beam space channel.Because of the limited main scattering paths of millimeter wave,the beam space channel is sparse.Under this system model,an adaptive sparsity support detection algorithm is proposed,in which the sparsity of channel does not need to be given.The simulation results indicate that compared with the traditional orthogonal matching pursuit algorithm,the proposed algorithm can reach higher accuracy at low signal-to-noise ratio.In the aspect of hybrid precoding,based on hybrid precoding structure composed of switches and inverters with low-energy consumption,this thesis proposes an ACE algorithm with relative weight.The simulation results show that the proposed algorithm can improve the sum-rate and energy efficiency of the system.