Channel Estimation Algorithm for RCF-mmMIMO System Based on Deep Learning
In response to the problems of high computational complexity and high pilot overhead in channel estimation for large-scale multi input multi output(CF-mmMIMO)systems without cellular millimeter waves,a depth learning-based channel estimation algorithm for reconfigurable intelligent surface(RIS)aided CF-MMMIMO(RCF-MMMIMO)systems is proposed.The algorithm introduces RIS to replace some millimeter wave access points(AP),increased millimeter wave coverage and signal strength under CF-mmMIMO,and to some extent,it solves the problem of high hardware cost due to the large number of millimeter wave AP in the absence of cellular architecture.Secondly,the existence of RIS cascaded channel mapping relationships for different groups of antennas on the same AP was proved,and the channel mapping was modeled using FCNN neural network,significantly reduces the high computational complexity and pilot overhead of channel estimation in the central processing unit of cellular free systems.Simulation experiments show that,channel estimation algorithms can effectively improve the accuracy of channel estimation with low computational complexity and low pilot overhead,significantly improve system transmission efficiency.
NETL
NSTL
万方数据
