Model based Beamspace Channel Estimation for Millimeter Wave Massive MIMO System

Abstract

Channel estimation concerning channel power leakage is investigated for millimeter wave (mmWave) massive MIMO system where the base station (BS) is equipped with uniform planar array (UPA). To model the channel leakage, the leakage effect of two dimensional beamspace channel is firstly analyzed. A theorem is proved to show that a four-point star (FS)-based model is better than the existing dual-crossing (DC)-based model. Then a FS-based beamspace channel estimation algorithm is proposed, where the initial selection of dominant entries is computed offline and then the selection is iteratively refined until the stop condition is satisfied. Simulation results verify the effectiveness of our work and show that the FS-based channel estimation algorithm outperforms the DC-based channel estimation algorithm.