Hybrid BD-GMD Precoding for Multiuser Millimeter-Wave Massive MIMO Systems

Wei WU  Danpu LIU  

IEICE TRANSACTIONS on Communications   Vol.E102-B   No.1   pp.63-75
Publication Date: 2019/01/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.2018EBP3001
Type of Manuscript: PAPER
Category: Fundamental Theories for Communications
millimeter wave,  massive MIMO,  hybrid precoding,  block diagonalization,  geometric mean decomposition,  

Full Text: PDF>>
Buy this Article

The potential for using millimeter-wave (mmWave) frequencies in future 5G wireless cellular communication systems has motivated the study of large-scale antenna arrays to achieve highly directional beamforming. However, the conventional fully digital beamforming (DBF) methods which require one radio frequency (RF) chain per antenna element are not viable for large-scale antenna arrays due to the high cost and large power consumption of high frequency RF chain components. Hybrid precoding can significantly reduce the number of required RF chains and relieve the huge power consumption in mmWave massive multiple-input multiple-output (MIMO) systems, thus attracting much interests from academic and industry. In this paper, we consider the downlink communication of a massive multiuser MIMO (MU-MIMO) system in the mmWave channel, and propose a low complexity hybrid block diagonal geometric mean decomposition (BD-GMD) scheme. More specially, a joint transmit-receive (Tx-Rx) analog beamforming with large-scale arrays is proposed to improve channel gain, and then a low-dimensional BD-GMD approach is implemented at the equivalent baseband channel to mitigate the inter-user interference and equalize different data streams of each user. With the help of successive interference cancellation (SIC) at the receiver, we can decompose each user's MIMO channel into parallel sub-channels with identical higher SNRs/SINRs, thus equal-rate coding can be applied across the sub-channels of each user. Finally, simulation results verify that the proposed hybrid BD-GMD precoding scheme outperforms existing conventional fully digital and hybrid precoding schemes and is able to achieve much better BER performance.