Channel estimation for FDD multi-user massive MIMO systems: a greedy approach based on user clustering

Obtaining channel state information (CSI) at the transmitter side is essential to take advantage of spectral and energy efficiency in massive multiple-input multiple-output systems. In particular, due to a large number of antennas at the base station (BS) side, the required pilots for downlink channel estimation and the following CSI feedback in the uplink path would be prohibitively large when the system employs frequency-division duplex protocol for data transmission. In this study, the authors propose a novel compressed sensing algorithm which exploits the commonly shared sparsity among nearby users to reduce the estimation error and the number of assigned pilots accordingly. Moreover, to gather users' channel with common sparsity, a clustering procedure is introduced, which groups active users located in the cell according to their mean angle of arrivals received by the uplink signals at the BS antenna array. After clustering users properly, the proposed algorithm can exploit shared support set existing in each cluster efficiently, thereby improving CSI estimation performance. Numerical results demonstrate that the clustering idea along with the proposed algorithm, outperforms other solutions, and is capable of approaching the performance bound when the transmit power is increased.