Low‐complexity hybrid precoding for multi‐user massive MIMO systems: a hybrid EGT/ZF approach. Issue 5 (9th March 2017)

• Record Type: Journal Article
• Authors: Hanif, Muhammad; Yang, Hong‐Chuan; Boudreau, Gary; Sich, Edward; Seyedmehdi, Hossein
• Is Part Of: IET communications. Volume 11:Issue 5(2017)
• Page Start: 765
• Page End: 771
• Publication Date: 2017-03-09
• Subjects: precoding -- MIMO communication -- array signal processing -- radiofrequency interference -- singular value decomposition; low complexity hybrid precoding -- multiuser massive MIMO systems -- hybrid EGT/ZF approach -- base station -- hybrid precoding design -- subconnected analogue -- analogue phased arrays -- multiuser massive MIMO system -- equal gain transmission -- EGT -- analogue beamforming -- analogue phased array -- employ zero‐forcing -- interuser interference -- singular value decomposition; Telecommunication systems -- Periodicals; Speech processing systems -- Periodicals; 621.38205
• DOI: 10.1049/iet-com.2016.0983
• Languages: English
• ISSNs: 1751-8628

Abstract

Abstract : Massive multiple‐input multiple‐output (MIMO) systems bring manifold improvements in the system spectral efficiency but result in high hardware and processing complexity at the base station. Employing hybrid precoding at the base station can reduce such complexity. In this study, unlike most existing work on hybrid precoding design, the authors consider a sub‐connected analogue combining structure to reduce the complexity. Starting from an important observation on the effect of sequentially designed analogue phased arrays on users' sum rate, the authors develop three low‐complexity hybrid precoding schemes for a multi‐user massive MIMO system. The proposed schemes apply equal gain transmission (EGT) based analogue beamforming to reap the diversity benefit of an analogue phased array and employ zero‐forcing (ZF) beamforming for nullifying inter‐user interference. The authors carry out an extensive computational complexity analysis and simulation study on the proposed schemes. The proposed singular‐value decomposition based EGT scheme outperforms all others but incurs the highest computational burden. On the other hand, the sequential‐EGT scheme is the least computationally intensive scheme but shows the worse performance amongst them.