Robust multi-user opportunistic beamforming for sparse networks

A scheme exploiting reduced feedback for the purpose of opportunistic multi-user beamforming is proposed. The scheme builds on recent promising advances realized in the area of multi-user downlink precoding and scheduling based on partial transmitter channel state information (CSIT) using random beamforming-based SDMA. Although random precoding followed by SDMA scheduling is optimal within the set of unitary precoders, it is only so in the asymptotic number of users K. For -practically relevant- sparse networks (i.e. with low to moderate number of users) random beamforming SDMA yields severely degraded performance. In this work we present a scheme allowing to restore robustness with respect to cell sparsity. The core idea here is to preserve the low complexity low feedback advantage of random opportunistic beamforming in selecting a target group of users, while much more efficient beamforming schemes can be used to serve the group of users once it has been identified. We propose different designs, optimal and suboptimal, based upon variable levels of feedback requirement. We show substantial gain over opportunistic beamforming for a range of K.