Contents I. Introduction
In massive MIMO systems, multi-user beamforming in dedicated channels has been investigated intensively [1], [2]. But designs for public channels are relatively limited. Public channels play crucial roles in broadcasting essential synchronization, reference, and control signals [3]. Public channel transmissions can be divided into two categories [4]: closed-loop and open-loop. For closed-loop approaches, instantaneous or statistic channel state information (CSI) is assumed at the BS and the transmission is optimized to be adaptive to the CSI, e.g., choosing the optimal precoding matrix to maximize the worst-case receiving signal-to-noise ratio (SNR) [5], [6]. Open-loop approaches assume no CSI and the BS broadcasts common information blindly to users. For some applications, quality CSI is difficult to obtain (e.g., users may be silent for a long time [7] ). Meanwhile, when the users’ channel covariance matrices are largely different, CSI-based designs have higher complexity and little performance superiority. Thus, this work considers open-loop schemes [3], [8].
