A least square channel estimator with pilot deficit for multiuser massive MIMO downlink

Abstract

The problem of channel estimation in multiuser massive multiple-input multiple-output (MIMO) downlink is considered. In multiuser MIMO systems, channel state information (CSI) is required for user scheduling and precoder design, and its erroneous estimation is detrimental to the system performance. As prior works have shown, in order to maintain system performance at a specific level, the number of pilots must increase proportional to the number of transmit antennas, which in massive MIMO systems is very difficult if not impossible to achieve. In this work, a situation where there are not enough pilots for channel estimation is considered. Computational shortcuts method for least square estimator is used, and using simulation results it is shown how extra antennas of massive MIMO can compensate the shortcut in number of pilots. As an example, it is shown that increasing number of base station (BS) antennas from 100 to 190 can compensate for pilot sequence length reduction of 50 to 25. The aforementioned trade-off shows how massive MIMO can be advantageous over fast fading channels, for which pilot sequence length must be kept short.