Hybrid Constraint-Aware Genetic Optimization for Structurally Coupled Resource Allocation, User Scheduling, and Grouping: An SBMA Case Study

This paper studies joint antenna allocation, subcarrier allocation, user activation, and superuser grouping in sparse-code-and-blind-interference-alignment-based multiple access (SBMA), where these decisions also determine the feasible transmission configuration. We formulate the problem as a constrained integer optimization and evaluate a hybrid constraint-aware genetic algorithm (HCAGA) using SBMA as a case study. HCAGA combines a compact integer encoding, feasibility-first selection, uniform crossover, multi-mode mutation, hierarchical structural and rate repair, and periodic QoS-aware local search. Its objective is to maximize the number of users whose rate meets a prescribed threshold. In 50 Monte Carlo realizations with 20 transmit antennas, 5 clusters, 20 subcarriers, and 20 candidate users per cluster, HCAGA serves 54.70 users at 10 dB and 93.54 at 20 dB on average, compared with 48.10 and 85.52 for a Standard GA and 25.00 for the tested budgeted random-search (BRS) implementation. The gain is operating-region dependent: Standard GA is significantly better at 6 dB, whereas the 4 dB difference is not significant and HCAGA is higher at most other tested SNRs. Paired ablations show positive average contributions from hierarchical repair and local search, with repair gains remaining statistically detectable at the tested 6, 12, and 18 dB points. HCAGA also requires substantially more rate evaluations; equal-evaluation-budget performance has not been established.