Channel Estimation for Hybrid mmWave MIMO Systems With CFO Uncertainties

Channel estimation at millimeter wave (mmWave) allows designing hybrid precoders and combiners to optimize performance metrics, such as the spectral efficiency. One major source of difficulty when estimating the high-dimensional and large-bandwidth channel, however, is the carrier frequency offset (CFO) that impairs the received signal. Most of the prior work assumes perfect time–frequency synchronization when estimating the channel. However, in practical systems, this assumption does not hold, and joint time–frequency synchronization is an important issue to deal with. In this paper, we propose a multi-stage algorithm for joint CFO and channel estimation. Using a training protocol between a transmitter and a receiver, we estimate first the equivalent low-SNR beamformed channel and the CFO, using the maximum-likelihood (ML) criterion. Thereafter, using these estimates, the high-dimensional multiple-input multiple-output (MIMO) channel is estimated leveraging its angular sparsity. The simulation results show that, using our proposed strategy, small estimation errors and near-optimal values of spectral efficiency can be achieved, without incurring in significant overhead and/or computational complexity.