AI-Aided Height Optimization for NOMA-UAV Networks

Abstract

Currently, Unmanned Aerial Vehicles (UAVs) are gaining significant attention due to their potential to effectively carry out a variety of tasks with superior performance through the use of fifth-generation (5G) and sixth-generation (6G) networks. Non-orthogonal multiple access (NOMA) techniques can further improve the performance and efficiency while reducing the interference. In this paper, we propose the application of machine learning (ML) techniques to evaluate the outage performance of a NOMA-enabled UAV network. Specifically, this study investigates the optimal UAV height that allows two users on the ground to receive the best service when they are simultaneously served by one UAV. We generated our own dataset which included several network parameters. We then trained various machine learning techniques on this dataset, including artificial neural networks (ANN), support vector regression (SVR), and linear regression (LR). Our results indicate that ANN provides the best accuracy compared with SVR and LR, with an average root mean squared error (RMSE) of 0.0931.