QoS-Driven Resource Allocation for Multiple Access in B5G and 6G Networks Using Metaheuristic Optimization

Resource allocation that is both energy-efficient and fair remains a persistent challenge in 5G, beyond-5G, and emerging 6G networks, where users with heterogeneous service requirements compete for limited radio resources. Although rate-splitting multiple access (RSMA) is commonly investigated through physical-layer models, service-level datasets often contain demand indicators without channel matrices, beamforming variables, or stream-level power parameters. We develop and evaluate a QoS-driven allocation framework based on three allocation-inspired models: an RSMA-inspired balanced model, a NOMA-inspired signal-strength-aware model, and an OFDMA proportional-fair reference model. Four metaheuristic optimizers—Genetic algorithm(GA), simulated annealing (SA), particle swarm optimization (PSO), and ant colony optimization (ACO)—are applied to the same QoS demand profile, resource budget, constraints, feasibility-repair procedure, and performance measures, producing 12 model–optimizer combinations. The input comprises 400 user records, including 393 records with positive bandwidth requirements and seven records with zero bandwidth. Each method was evaluated over 30 independent runs. Under this QoS-level abstraction, RSMA-PSO produced the highest mean fitness and energy-efficiency values, whereas RSMA-GA achieved the strongest composite balance across fitness, energy efficiency, fairness, and satisfaction. The mean fitness of RSMA-GA was 9.32% higher than that of NOMA-GA and 18.00% higher than that of OFDMA-PF-GA. These results characterize allocation behavior under a shared QoS profile and should not be interpreted as full physical-layer validation of RSMA, NOMA, or OFDMA transmission.