Machine Learning-Based User Scheduling in Integrated Satellite-HAPS-Ground Networks

Integrated space-air-ground networks promise to offer a valuable solution space for empowering next generation of communication networks (6G), particularly in the context of connecting the unconnected and ultraconnecting the connected. Such digital inclusion thrive makes the resource management problem of particular interest. However, the classical model-based optimization methods cannot meet the real-time processing and user's QoS needs, due to the high heterogeneity of the space-air-ground networks and the complexity of its associated resource allocation problems. Given the premises of artificial intelligence at automating wireless networks design, this paper focuses on showcasing the prospects of machine learning in the context of user scheduling in integrated space-air-ground communications. The paper first overviews the most relevant state-of-the art in the context of machine learning applications to the resource allocation problems in integrated space-air-ground networks. The paper then proposes, and shows the benefit of, one specific use-case that adopts ensembling deep neural network for optimizing the user scheduling policies in space-high altitude platform station (HAPS)-ground networks. Finally, the paper presents some challenges and sheds light on several open issues in the context of machine learning applications in space-air-ground networks, namely, power limit, imperfect channel state information, multi-HAPSs scenarios and flying taxis-empowered systems.