A Machine Learning-Based Uplink Resource Allocation Technique for Mixed Traffic in Non-Terrestrial Networks

Abstract

The integration of unmanned aerial vehicles (UAVs) and low earth orbit (LEO) satellites in communication networks is receiving increasing attention pursued in current research efforts. The UAVs, with links to LEO satellites, can be used as base stations to provide coverage in remote out-of-coverage areas. This paper proposes a machine learning (ML) based deployment of a single UAV that allocates uplink resources to cover mixed traffic demands. The UAV is positioned in the 3-D space and the resources are allocated to fulfill the requirements of all user equipment (UE) traffic profiles. These traffic profiles include the Ultra-reliable and low-latency communication (URLLC), the enhanced Mobile Broadband (eMBB) and the Age of Information (AoI) sensitive traffic. Our proposed technique aims at obtaining real-time results close to the optimal bound of the solution at lower complexity. Simulation results show that the proposed solution achieves close-to-optimal results and outperforms benchmark techniques from previous studies.