Year: 2026
Type: article
Abstract: This study revisits the joint design of user clustering, Access-Point (AP) association, transmit-antenna selection, and power allocation in downlink cell-free massive MIMO with Non-Orthogonal Multiple Access (NOMA). A rigorous Monte-Carlo evaluator with pilot contamination, Minimum Mean-Square-Error (MMSE) channel estimation, maximum-ratio transmission, and Successive Interference Cancellation (SIC) scores every candidate configuration consistently under common random numbers. A central methodological point is emphasized: prior clustering-centric comparisons hold AP association and power fixed, which conflates the clustering effect with an unoptimized association budget. When all schemes are granted an identical association/antenna/power search-a matched-scope comparison over 40 independent topologies—a clear and statistically significant ordering of resource levers emerges: AP association is dominant (+20.2 bps/Hz), clustering optimization is a significant second lever (+13.0 bps/Hz), the clustering algorithm itself contributes a small but significant increment (+4.1 bps/Hz), and transmit-antenna selection is spectrally neutral, all under the parameter regime investigated (Table 3); the ordering is not asserted to hold universally across every cell-free massive MIMO-NOMA configuration. A block-coordinate joint optimizer is proposed and shown, by exhaustive enumeration on small instances (K=9), to attain a mean optimality gap of only 0.64%, a result that is evidence for, but not proof of, near-optimality at the larger K=30 scale used elsewhere in the evaluation; it improves the sum Spectral Efficiency (SE) by 16.0% over the best matched-scope heuristic (and by 54.3% over the conventionally evaluated base method) while converging monotonically in about five iterations. Its discrete power grid is shown to attain 99.3% of a continuous power optimum. Transmit-antenna selection is characterized as an energy-efficiency lever, the energy-optimal configuration activating four of eight antennas for a 14.3% energy-efficiency gain at a 28.2% SE cost.
Authors: Gurudevi A. Hiremath, Shankar D. Nawale
Institution: Solapur University
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7225909651
