A Physics-Based 6G OAM Terahertz Dataset: A Foundation for Machine Learning and Deep Reinforcement Learning in Next-Generation Wireless Networks

This letter outlines a holistic, physics-based dataset to accelerate research into sixth-generation (6G) wireless networks. This dataset for Terahertz (THz) communications is based around multiplexing approaches with Orbital Angular Momentum (OAM) and consists of 270,000 samples with 33 parameters, all sourced from a high-fidelity channel model based on the ITU-R IMT-2030 framework and validated on a common machine learning task for predicting throughput, resulting in a coefficient of determination R²=0.92. By combining theoretical models with data-driven techniques, this work will advance the design, development, and validation of state-of-the-art ML and DRL algorithms for next-generation wireless communication systems.