ICA Filtering Based Channel Estimation for Massive MIMO TDD Systems

A novel ICA-based channel estimation method for massive MIMO systems has been proposed in this letter. The proposed method first employs the systems' very large transmit antennas to obtain channel time-delay vectors through classical ICA schemes, and then employs the independent and slowly time-varying time-delay vectors to filter the pilot contamination in the least-squares channel estimate during the time-delay vectors' coherence interval, which is much longer than the channel's coherence interval. The proposed method can efficiently suppress the pilot contamination without increasing the system pilot overhead. The simulation results demonstrate the effectiveness of the proposed method.