AMSA — Adaptive ML-based Synchronization Arbiter: training dataset and code

Reproducibility package for the doctoral dissertation on AMSA — an Adaptive ML-based Synchronization Arbiter for time-synchronization source selection in 5G/6G transport networks. AMSA replaces the deterministic BMCA (IEEE 1588) with a Random Forest classifier using time-error statistics and a contextual-memory feature. This package contains the training datasets (simulation-generated, including train_ready.csv), the data-generation and model-training scripts (random_state = 42), and the trained model (model.pkl). See README.md for structure and reproduction steps.