Research On Channel Estimation Method Based On Matrix Sensing For Millimeter Wave MIMO Systems

Millimeter wave(mm Wave)technology has been regarded as a key technology for 5G.This frequency band is rich in spectrum resources and can offer orders of magnitude greater bandwidth than the current cellular bands.Using millimeter wave frequency bands for communication can effectively alleviate the problem of spectrum scarcity.The combination of millimeter wave technology and massive MIMO technology can greatly improve the system capacity.Channel state information is useful for designing efficient hybrid MIMO processors of mm Wave systems.Channel state information can be obtained through channel estimation.Channel estimation for hybrid millimeter wave massive MIMO systems is challenging because the number of radio frequency chains is smaller than the number of antennas.Because of the short wavelength,fast fading and the use of hybrid beamforming,the channel estimation technology of the millimeter wave system needs to be redesigned.Traditional channel estimation usually uses compressive sensing algorithms.The methods based on compressive sensing require the knowledge of the array response,which may suffer performance degradation due to unknown phase errors and gain errors caused by the array-inherent impairments.This thesis designs a channel estimation scheme based on matrix sensing for large-scale mm Wave MIMO systems with hybrid transceiver structures based on phase shifters.Matrix sensing-based channel estimation exploits the low-rank nature of mm-wave channels.This scheme is robust to the arrayinherent impairments.The RTr DC algorithm with the best performance is selected by comparing the existing matrix sensing algorithms.In order to improve the performance and speed up the algorithm iteration,this paper proposes an improved NRTr DC algorithm based on RTr DC.The proposed algorithm can recover the mm Wave channel,and is not affected by array-inherent impairments.The simulation results show that the proposed NRTr DC algorithm has better performance on the NMSE.The proposed method is not affected by array response errors,and can obtain near-optimal SE.In summary,channel estimators based on matrix sensing provide a channel estimation scheme with better performance and robustness,and provides a meaningful research direction for subsequent mm Wave channel estimation.