Optimizing Multi-User Uplink Cooperative Rate-Splitting Multiple Access: Efficient User Pairing and Resource Allocation with Gradient-based Meta Learning

• Published in: IEEE transactions on communications p. 1
• Main Authors: Khisa, Shreya, Elhattab, Mohamed, Assi, Chadi, Sharafeddine, Sanaa
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: Cooperative communications; Decoding; Downlink; half-duplex; Interference; meta learning; NOMA; Optimization; Resource management; RSMA; Throughput; Training; Uplink; user pairing; Wireless communication
• ISSN: 0090-6778; 1558-0857
• DOI: 10.1109/TCOMM.2025.3550371

This paper investigates joint user pairing, power, and time slot duration allocation in the uplink multiple-input single-output (MISO) multi-user cooperative rate-splitting multiple access (C-RSMA) networks in half-duplex (HD) mode. We assume two types of users: cell-center users (CCU) and cell-edge users (CEU); first, we propose a user pairing scheme utilizing a semi-orthogonal user selection (SUS) and a matching-game (MG)-based approach where the SUS algorithm is used to select CCU in each pair. Afterward, the CEU in each pair is selected by considering the highest channel gain between CCU and CEU. After pairing is performed, the communication occurs in two phases: in the first phase, in a given pair, CEUs broadcast their signal, which is received by the base station (BS) and CCUs. In the second phase, in a given pair, the CCU decodes the signal from its paired CEU, superimposes its own signal, and transmits it to the BS. Moreover, utilizing uplink RSMA principle, only the CCUs split their messages into two sub-messages. Meanwhile, the messages of CEUs are kept without splitting. We formulate a joint optimization problem in order to maximize the sum rate subject to the power budget constraints of the user equipment (UE) and minimum data rate requirements at each UE. Since the formulated optimization problem is non-convex, we adopt a bi-level optimization to make the problem tractable. We decompose the original problem into two sub-problems: the user pairing sub-problem and the resource allocation sub-problem, where the user pairing sub-problem is independent of the resource allocation sub-problem, and once pairs are identified, the resource allocation sub-problem is solved for a given pair. The resource allocation sub-problem is solved by invoking a low-complexity pre-training free gradient-based meta-learning (GML) algorithm. Simulation results demonstrate that our proposed C-RSMA scheme can achieve around 100%, 51%, 53%, and 215% improvement over C-NOMA with fixed time slot allocation, RSMA, NOMA, and C-RSMA random pairing, respectively at CEU power budget of 17 dBm.