STAR-RIS-Assisted Hybrid MIMO mmWave Communications

• Published in: IEEE internet of things journal Vol. 11; no. 21; pp. 35141 - 35154
• Main Authors: Yang, Xiawei, Liu, Heng, Gong, Shiqi, Wang, Gongpu, Xing, Chengwen
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.11.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Array signal processing; Beamforming; Binary quadratic programming (BQP); Complexity; Constraints; Coupled modes; Finite element analysis; hybrid analog-digital precoder/combiners; Mathematical programming; Millimeter wave communication; Millimeter waves; Optimization; Phase shift; Protocol; Protocols; Quadratic programming; Reconfigurable intelligent surfaces; simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS); Vectors; Wireless networks
• ISSN: 2327-4662; 2327-4662
• DOI: 10.1109/JIOT.2024.3436831

The simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) has been a promising enabler for the future wireless network due to its full-space coverage capability. In this article, we investigate the STAR-RIS assisted hybrid mmWave multiple-input-multiple-output system, where the two practical operating protocols, i.e., energy splitting (ES) and mode switching (MS), and the coupled transmission and reflection (T&R) phase-shift model for the STAR-RIS are considered. For each operating protocol, we aim to maximize the system weighted sum rate (WSR) by jointly optimizing the passive T&R coefficients at the STAR-RIS and the hybrid analog-digital precoder/combiners, subject to the discrete phase shift constraints. Specifically, we propose an efficient weighted minimum mean-square error based alternating optimization (AO) algorithm to address this highly coupled nonconvex problem. By leveraging the special ordered set of type 1 under the MS protocol, the optimization of both the discrete T&R coefficients and analog precoder/combiners can be equivalently transformed into the standard binary quadratic programming, which can be effectively solved by the mathematical programming with the equilibrium constraints-based exact penalty algorithm. The proposed penalty-based AO algorithm is also applicable to the WSR maximization under the ES protocol. In addition, to avoid high-complexity iterative process wherever possible, we develop a separate analog-digital beamforming scheme, where a fast projection-based gradient descent algorithm is applied to successively optimize discrete T&R coefficients and analog precoder/combiners to maximize the effective channel gain, and then the optimal digital precoder/combiners are obtained in semi-closed forms. Numerical simulation results demonstrate the superior WSR performance and complexity advantage of the proposed algorithms over the existing benchmark schemes.