Joint CSIT Acquisition Based on Low-Rank Matrix Completion for FDD Massive MIMO Systems

摘要

Channel state information at the transmitter (CSIT) is essential forfrequency-division duplexing (FDD) massive MIMO systems, but conventionalsolutions involve overwhelming overhead both for downlink channel training anduplink channel feedback. In this letter, we propose a joint CSIT acquisitionscheme to reduce the overhead. Particularly, unlike conventional schemes whereeach user individually estimates its own channel and then feed it back to thebase station (BS), we propose that all scheduled users directly feed back thepilot observation to the BS, and then joint CSIT recovery can be realized atthe BS. We further formulate the joint CSIT recovery problem as a low-rankmatrix completion problem by utilizing the low-rank property of the massiveMIMO channel matrix, which is caused by the correlation among users. Finally,we propose a hybrid low-rank matrix completion algorithm based on the singularvalue projection to solve this problem. Simulations demonstrate that theproposed scheme can provide accurate CSIT with lower overhead than conventionalschemes.