HAPS-UAV-Enabled Heterogeneous Networks: A Deep Reinforcement Learning Approach

The integrated use of non-terrestrial network (NTN) entities such as the high-altitude platform station (HAPS) and low-altitude (LAPS) has become essential elements in space-air-ground networks (SAGINs). However, complexity, mobility, heterogeneity NTN resources present various challenges from system design to deployment. This paper proposes a novel approach designing heterogeneous consisting HAPSs unmanned aerial vehicles (UAVs) being LAPS entities. Our involves jointly optimizing three-dimensional trajectory channel allocation for base stations, with focus on ensuring fairness provision quality service (QoS) ground users. Furthermore, we consider load stations incorporate this information into optimization problem. proposed utilizes combination deep reinforcement learning fixed-point iteration techniques determine UAV locations strategies. Simulation results reveal that our learning-based significantly outperforms conventional benchmark models.