Integrated space-air-ground networks promise to offer a valuable solution space for empowering the sixth generation of communication networks (6G), particularly in the context of connecting the unconnected and ultraconnecting the connected. Such digital inclusion thrive makes resource management problems, especially those accounting for load-balancing considerations, of particular interest. The conventional model-based optimization methods, however, often fail to meet the real-time processing and quality-of-service needs, due to the high heterogeneity of the space-air-ground networks, and the typical complexity of the classical algorithms. Given the premises of artificial intelligence at automating wireless networks design and the large-scale heterogeneity of non-terrestrial networks, 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, with a dedicated attention to space-air-ground networks. The paper then proposes, and shows the benefit of, one specific use case that uses ensembling deep neural networks for optimizing the user scheduling policies in integrated space-high altitude platform station (HAPS)-ground networks. Finally, the paper sheds light on the challenges and open issues that promise to spur the integration of machine learning in space-air-ground networks, namely, online HAPS power adaptation, learning-based channel sensing, data-driven multi-HAPSs resource management, and intelligent flying taxis-empowered systems.
翻译:综合空地网络有望为赋予第六代通信网络(6G)赋权提供宝贵的解决方案空间,特别是在连接未连接和超链接的通信网络(6G)的背景下。这种数字包容的兴盛使得资源管理问题,特别是考虑负载平衡因素的问题特别令人感兴趣。常规模型优化方法往往无法满足实时处理和服务质量的需要,原因是空空地网络高度不均匀,以及经典算法的典型复杂性。鉴于无线网络设计自动化的人工智能以及非地铁网络的大规模异质性,本文件侧重于展示在综合空地通信用户时间安排方面机器学习的前景。由于空间空地网络和传统算法的典型复杂性很大,本文件首先概述了机器学习应用对资源分配问题最为相关的艺术,并专门关注空地网络。鉴于无线网络设计自动化的人工智能以及非地铁网络的大规模异质性,本文件侧重于展示机器学习的前景,从而优化空间-空空空间-陆空间-空间-空间-空间-轨道-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-高级-智能-智能-智能-智能-智能-系统平台平台平台平台平台平台-优化系统-智能-智能-智能-智能-智能-智能-智能-智能-智能、智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能-智能定位-智能-智能-智能-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位-定位