Reconfigurable intelligent surface (RIS)-assisted aerial non-terrestrial networks (NTNs) offer a promising paradigm for enhancing wireless communications in the era of 6G and beyond. By integrating RIS with aerial platforms such as unmanned aerial vehicles (UAVs) and high-altitude platforms (HAPs), these networks can intelligently control signal propagation, extending coverage, improving capacity, and enhancing link reliability. This article explores the application of deep reinforcement learning (DRL) as a powerful tool for optimizing RIS-assisted aerial NTNs. We focus on hybrid proximal policy optimization (H-PPO), a robust DRL algorithm well-suited for handling the complex, hybrid action spaces inherent in these networks. Through a case study of an aerial RIS (ARIS)-aided coordinated multi-point non-orthogonal multiple access (CoMP-NOMA) network, we demonstrate how H-PPO can effectively optimize the system and maximize the sum rate while adhering to system constraints. Finally, we discuss key challenges and promising research directions for DRL-powered RIS-assisted aerial NTNs, highlighting their potential to transform next-generation wireless networks.
@article{arxiv.2412.18957,
title = {RIS-Assisted Aerial Non-Terrestrial Networks: An Intelligent Synergy with Deep Reinforcement Learning},
author = {Muhammad Umer and Muhammad Ahmed Mohsin and Aryan Kaushik and Qurrat-ul-Ain Nadeem and Ali Arshad Nasir and Syed Ali Hassan},
journal= {arXiv preprint arXiv:2412.18957},
year = {2024}
}