Accurate Channel Estimation for mmWave Massive MIMO With Partially Coherent Phase Offsets

This letter studies how to accurately estimate channels for millimeter wave (mmWave) massive MIMO in presence of partially coherent phase offsets. This is motivated by the observation that the received pilots share almost the same phase offset in the same training frame, but have different phase offsets across different training frames. In consideration of this property, we propose a two-stage estimation scheme to jointly estimate channels and phase offsets for mmWave massive MIMO systems. Firstly, based on an on- grid signal model, we use expectation-maximization generalized approximate message passing (EM-GAMP) algorithm to estimate the sparse angle-domain channel and phase offsets. Secondly, knowing the on- grid estimation results, we recast the formulation so as to reduce the number of unknown channel parameters to be refined. Then an EM-based algorithm is proposed to refine the channel and phase offsets estimates. Simulation results reveal substantial performance improvement over the existing counterparts.