Research On Channel Estimation With Compressive Sensing For Massive MIMO

Massive multiple-input multiple-output(M-MIMO)is becoming a key technology for future 5G wireless communications.In MIMO systems,channel state information(CSI)is essential for both space-time coding at transmitters and signal detection at receivers.However,in wireless communication,the propagation environment of the signal varies with time,and CSI obtained at the base station will differs from the actual channel because of delay,which will make communication quality deteriorate.To overcome this problem,channel estimation(CE)is required.Channel estimation technology has become a common technology in wireless communication,but because of the huge number of antennas in M-MIMO system,the channel estimation technology has been challenged.In recent years,the development and application of Compressive Sensing(CS)theory has provided a new approach for channel estimation.From a spatial point of view,the adjacent channels of a large-scale MIMO system have strong correlation in space.In practice,the energy of the wireless channel is concentrated in a limited spatial direction,that is,the channel can be compressed.From the frequency domain perspective,the frequency domain channel matrix is sparse based on the Fast Fourier Transform(FFT)matrix when the number of subcarriers of the system is much larger than the channel multipath number.The sparse channel response suggests that compression sensing can be introduced into the channel estimation,which greatly reduces the amount of channel response data that needs to be estimated and transmitted.In this thesis,we consider the downlink channel estimation training in a massive MIMO system,where the base station is equipped with a large-scale planar array and each user equipment has a single antenna.The modified orthogonal matching pursuit(OMP)algorithm is proposed to estimate angular information of the dominant channel paths,which can be subsequently used for downlink beamforming in the data transmission mode.The modified OMP algorithm is adaptive.Capitalizing on compressive sensing,the idea of the proposed method is to refine the measurement matrix(beam directions)during the CSI acquisition process by exploiting the knowledge of the visibility region of a planar array.The simulation results show that the proposed algorithm can obtain better performance and improve the spectral efficiency compared with the classical OMP algorithm.On the other hand,channel feedback for massive MIMO is challenging due to the substantially increased dimension of MIMO channel matrix.For this reason,on the basis of the study of channel impulse response(CIR)feedback for massive MIMO systems,a segmented CIRs feedback scheme based on compressive sensing has been proposed.Specifically,segmented channels are sparser than the original channel.Thus,the base station can recover the highly compressed segmented CIRs under the framework of compressive sensing.Simulation results show that the proposed scheme can reduce the feedback error compared with the direct CS-based scheme and that when compression ratio is 20%,the direct CS-based scheme fails to work since the feedback while the proposed scheme performs well;when compression ratio is 50%,the proposed scheme achieves a 5 dB SNR gain compared with the direct CS-based scheme.