User Selection Algorithms for Block Diagonalization Aided Multiuser Downlink mm-Wave Communication

In order to meet the ever increasing data rate demands, the next generation of wireless communication systems is being designed for exploiting the large amounts of unused spectrum in the millimeter-wave (mm-wave) band. Since operating at mm-wave frequencies imposes several challenges, such as high-path loss, as well as both spatial and temporal channel sparsity, there is a significant research interest focused on designing feasible solutions for establishing reliable and high-throughput links at mmwave frequencies. In this paper, we consider a cellular system relying on hybrid beamforming aided base station (BS) as well as user equipment and study the user selection problem, which has not been hitherto studied in the literature. More specifically, we study the problem of selecting K'-users by the BS for communication out of K-users whilst ensuring that the sum rate is maximized. Specifically, we propose a user selection algorithm, which relies on the knowledge of both the channel gains and the angle of departure (AoD) of the channel paths spanning to the various users, which is termed the AoD aided user selection (AoD-US). Furthermore, we devise a pair of subspace metrics based on: 1) the angle between the subspaces spanned by the BS array response vectors; 2) the ratio of interference and signal space dimensions of various users, in order to reduce the user search space in AoD-US. This modified user selection algorithm is termed the AoD aided user selection with user set pruning (AoD-US-P). Furthermore, we study the attainable sum-rate performance of the block-diagonalization aided downlink and show that the proposed selection algorithms guarantee both multiuser diversity and multiplexing gains. Additionally, the proposed algorithms are studied in the round-robin scheduling scenario, where all the K-users are scheduled for achieving fairness. Our simulation results revealed that the AoD-US-P achieves nearly the same performance as that achieved by the AoD-US despite having a small user set, while both are observed to outperform the channel power-based selection scheme.