Hybrid Transceivers Design for Large-Scale Antenna Arrays Using Majorization-Minimization Algorithms

• Published in: IEEE transactions on signal processing Vol. 68; pp. 701 - 714
• Main Authors: Arora, Aakash, Tsinos, Christos G., Rao, Bhavani Shankar Mysore Rama, Chatzinotas, Symeon, Ottersten, Bjorn
• Format: Journal Article
• Language: English
• Published: New York IEEE 2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; alternating minimization; Antenna arrays; Antenna design; Antennas; Computer simulation; Convergence; Hybrid precoding; large-scale antenna arrays; majorization-minimization; Millimeter waves; MIMO (control systems); MIMO communication; mmWave MIMO communications; Optimization; Phase shifters; Power consumption; Precoding; Radio frequency; Signal processing algorithms; Transceivers
• ISSN: 1053-587X; 1941-0476; 1941-0476
• DOI: 10.1109/TSP.2019.2957611

Hybrid analog-digital (A/D) transceivers are an appealing solution to reduce the transceiver hardware complexity and power consumption for the millimeter wave (mmWave) communication and more general large-scale antenna array (LSAA) systems. In contrast to fully digital conventional multiple-input-multiple-output (MIMO) systems, the baseband precoding operation splits into a lower-dimensional digital precoder followed by a network of analog phase shifters. In this paper, we consider the hybrid precoder design as a constant modulus constrained matrix factorization (CMCMF) problem for the most common types of hybrid architectures namely, the fully and the partially connected ones. Two lines of algorithms based on the majorization-minimization (MM) and the minorization-maximization framework, respectively are proposed for these architectures. In particular, we present efficient algorithms scalable for LSAA systems with provable convergence guarantees to a stationary point. We also consider the hybrid postcoder design at the receiver end. Simulation results demonstrate that the proposed algorithms converge faster to a stationary point as compared to the state-of-the-art solutions that exist in literature. Furthermore, the solution tailored for the partially connected case achieves significantly improved performance in terms of the system spectral efficiency when compared to the existing solutions.