Downlink Channel Estimation for FDD Massive MIMO OFDM Systems

It is challenging for the evolved Node B (eNB) to obtain downlink channel state information (CSI) for frequency-division duplex (FDD) massive multiple-input multiple-output (MIMO) systems. To address the downlink channel estimation problem for FDD massive MIMO OFDM systems, this paper proposes a practical method by exploiting the reciprocal properties between the uplink and the downlink channels. Specifically, in uplink, the eNB estimates the reciprocal channel parameters such as the path number, path delays, path amplitudes, and path angles by mainly using compressive sensing (CS) techniques and QR-factorization. And then in downlink, the users estimate the non-reciprocal channel parameters, i.e., the path phases. After the users feed back the path-phase estimates to the eNB, the downlink CSI can be reconstructed at the eNB. Simulation results demonstrate that compared with the existing method, the proposed approach can exhibit better estimation performance with less computational complexity.