A new RF beam training method for multi-user millimeter wave systems

When implementing single radio frequency (RF) chain multi-user (MU) millimeter-wave (mmWave) systems, an RF beamforming algorithm with a short beam training overhead is essential. In this paper, we propose a new MU RF beamforming algorithm based on the conventional RF beam training method in IEEE 802.11ad. Then, we investigate its asymptotic behavior for a large number of users scenario. We show that the proposed scheme approaches the optimal full search scheme with much reduced beam training latency as the number of users grows. Our simulation results demonstrate that a performance gain of the proposed method over the full search method is about 35% in practical MU mmWave environments in terms of the effective data rate.