Beam Structured Channel Estimation for Massive MIMO with Beam Completion

In this paper, we investigate the channel estimation for massive multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems. By utilizing the beam based statistical channel model (BSCM), the aim is to estimate the beam domain channel and then convert it to the space-frequency (SF) domain channel. The problem is solved by the EM algorithm based on beam completion with information geometry theory. By completing the measurement matrix to a unitary matrix, we can get the equivalent received signal model, and then the model manifold and data manifold related to the a posteriori distribution of channel are built separately. The optimal distribution is obtained by <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$m$</tex>-projecting and m-projecting alternately between these two manifolds. We demonstrate that the EM algorithm based on beam completion can converge to MMSE estimator with low complexity. Simulation results show that the proposed algorithm can achieve better performance than existing methods when the number of iterations is small.