An early research prototype for MU-MIMO user scheduling using a Gym-style environment, CNN-based policies, clustering, and conventional scheduling baselines.
This repository is retained as a historical prototype. Newer paper-oriented MU-MIMO AI-RAN work is maintained separately. Results here should not be treated as current paper evidence without revalidation.
- simulated MU-MIMO channel and rate evaluation;
- user selection and scheduling experiments;
- CNN-based scheduling policies;
- clustering-assisted candidate grouping;
- conventional scheduling comparisons.
Before reusing an experiment, verify its channel assumptions, antenna configuration, candidate-user count, scheduling limit, power setting, data split, and metric definition.