| The design of 6G will be based on the various technologies in the future,such as millimeter wave,massive multiple-input multiple-output(MIMO)and intelligent reflecting surface(IRS),so as to improve the capacity of system,efficiency of spectrum and reliability of transmission.To give full play to the potential performance of the system,it is necessary to obtain complete and accurate channel state information(CSI).However,with the more complex of system composition,it becomes difficult to estimate CSI.Therefore,the channel estimation of two different scenarios is researched in this thesis,including millimeter wave massive MIMO system and IRS assisted millimeter wave massive MIMO system.The specific research contents are as follows:Firstly,the traditional channel estimation approaches applied for millimeter wave massive MIMO system involve overwhelming training overhead and poor performance.In this thesis,the problem of channel estimation is formulated as a low-rank matrix completion problem by utilizing the low rank structure of the millimeter wave massive MIMO channel matrix.An approach of channel estimation based on Schatten-p quasi-norm is called iterative weighted least squares(IRLS),which is used to recovery the channel matrix by utilizing incomplete CSI.In addition,the performance of the proposed approach is researched in the low rank ill conditioned channel environment.Simulations demonstrate that the proposed approach can solve the problem of overwhelming training overhead effectively,improve the accuracy of channel estimation and has stable estimation performance in high-dimensional and ill-conditioned channel environment.Secondly,the IRS is usually used to provide indirect communication services and solve the blocking problem of direct communication link between base station(BS)and mobile station(MS).However,the joint of active and passive beamforming technology and the the optimal control of IRS have higher standards for the accuracy of CSI in one IRS assisted millimeter wave massive MIMO system.Therefore,the channel estimation of reflection link channel is researched in this system.In this thesis,the reflection link channel is equivalent to the cascaded channel and the problem of channel estimation is formulated as a low-rank matrix completion problem based on matrix decomposition by utilizing the low rank structure of the cascaded channel matrix.An approach of channel estimation based on scaled gradient descent(SGD)framework is researched.Considering that the proposed approach requires the rank of the channel matrix as a priori information,and the rank is generally unknown,this thesis propose a strategy to adjust the rank of the channel matrix by eigenvalue of factor matrix.Simulations demonstrate that the proposed approach has lower computational complexity and higher estimation accuracy than other previous approaches. |
