Parallelizable First-Order Fast Algorithm for Symbol-Level Precoding in Lage-Scale Systems

• Published in: IEEE Vehicular Technology Conference pp. 1 - 6
• Main Authors: Yang, Junwen, Li, Ang, Liao, Xuewen, Masouros, Christos
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2023
• Subjects: ADMM; Approximation algorithms; constructive interference; Downlink; Interference; Jacobian matrices; Massive MU-MISO; parallel and distributed computing; Precoding; separability; Simulation; symbol-level precoding; Vehicular and wireless technologies
• ISSN: 2577-2465
• DOI: 10.1109/VTC2023-Spring57618.2023.10199658

We investigate constructive interference (CI)-based symbol-level precoding (SLP) in large-scale systems with massive connectivity of users to minimize the transmit power subject to the instantaneous signal-to-interference-plus-noise-ratio (SINR) and CI constraints. By converting the considered problem into a novel separable formulation, we reveal the existence of separability in SLP, which is therefore well-suited for decomposition. The proximal Jacobian alternating direction method of multipliers (PJ-ADMM) framework is adopted to decompose the reformulated problem into multiple subproblems, which can be solved in parallel with closed-form solutions. We further linearize the second-order terms by approximation, which leads to a parallelizable first-order fast solution to SLP. Our derivations are validated by simulation results, which also show that our algorithm can provide optimal performance with substantially lower computational complexity than state-of-the-art algorithms.