An Efficient Downlink Channel Estimation Approach for TDD Massive MIMO Systems

In this paper, channel estimation problem for downlink massive multi-input multi-output (MIMO) system is considered. Motivated by the observation that channels in massive MIMO systems may exhibit sparsity and the path delays vary slowly in one uplink-downlink process even though the path gains may be quite different, we propose a novel channel estimation method based on the compressive sensing. Unlike the conventional methods which do not make use of any a priori information, we estimate the probabilities that the paths are nonzero in the downlink channel by exploiting the channel impulse response (CIR) estimated from the uplink channel estimation. Based on these probabilities, we propose the Weighted Structured Subspace Pursuit (WSSP) algorithm to efficiently reconstruct the massive MIMO channel. Simulation results show that the WSSP could reduce the pilots number significantly while maintain decent channel estimation performance.