Hybrid Analog and Digital Precoding Design for Minimum BER in Massive MIMO System

• Published in: IEEE transactions on vehicular technology Vol. 73; no. 7; pp. 10060 - 10074
• Main Authors: Li, Junjie, Zhao, Lei, Jiang, Yuan
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Alternative energy sources; BER minimization; Complexity; Constraints; coordinate descent; Design optimization; Error analysis; hybrid analog and digital precoding; Hybrid systems; Lower bounds; manifold optimization; Massive MIMO; Minimization; Optimization; Phase shifters; Precoding; Radio frequency; Signal to noise ratio; Streams
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2024.3366233

Hybrid analog and digital precoding technology has been widely investigated recently in massive multiple-input multiple-output (mMIMO) systems as low-complexity and low-cost alternatives to fully digital precoding. Most existing hybrid precoding studies aim to improve the system's performance in terms of spectral efficiency (SE), energy efficiency (EE), or mean square error (MSE), which may have limited BER performance. To improve the system's transmission reliability, this paper investigates the design of the hybrid precoding/combining for minimum BER in the single-user mMIMO system. First, we formulate an average BER minimization optimization problem, subject to the maximum transmit power constraint and the constant modulus constraints for the phase shifters. Further, we equivalently transform the BER minimization problem into a constrained MMSE problem to make it tractable and consequently derive the closed-form digital precoder that can achieve the derived lower bound of the average BER. Finally, we propose an alternating optimization procedure (AOP) to jointly design the analog precoder and combiner, where a high-performance manifold optimization (MO)-based algorithm and a low-complexity coordinate descent (CD)-based algorithm are presented to overcome the difficulty incurred by the constant modulus constraints. Simulation results demonstrate that the proposed minimum BER hybrid precoding significantly outperforms the existing hybrid precoding in terms of BER and achieves the same level as the benchmark schemes in terms of SE.