Combined with artificial intelligence (AI) technology, Space-Air-Ground Integrated Networks (SAGINs) play a crucial role in realizing the 6G vision of self-awareness, ubiquitous intelligence, and Internet of Everything (IoE). Compared with 5G, the 6G vision demands higher performance in key performance indexes (KPIs) such as peak data rate, user experience data rate, delay, coverage percentage, reliability, etc. And, the independent configuration and deployment of network functions through network deployment automation is essential for meeting these 6G KPIs. However, traditional deployment strategies lack flexibility and applicability, relying on manual intervention. To address this, we analyze the characteristics of various AI algorithms in 6G SAGINs and propose a federated learning (FL)-assisted deep reinforcement learning (DRL) framework, which jointly optimizes deployment strategies through local and global collaboration. Case studies verify the effectiveness of this approach in improving network deployment automation and ensuring related KPIs in data management, resource allocation, and other tasks. Finally, we discuss the significant challenges that AI will face in deploying 6G SAGIN settings.

中文翻译：

---

6G 天地一体化网络的人工智能部署自动化：挑战、设计和展望


结合人工智能（AI）技术，天地空地一体化网络（SAGIN）在实现自我感知、泛在智能和万物互联的6G愿景中发挥着至关重要的作用。与5G相比，6G愿景对峰值数据速率、用户体验数据速率、时延、覆盖率、可靠性等关键性能指标（KPI）要求更高，并且通过网络部署实现网络功能的独立配置和部署自动化对于满足这些 6G KPI 至关重要。然而，传统的部署策略缺乏灵活性和适用性，依赖于人工干预。为了解决这个问题，我们分析了6G SAGIN中各种人工智能算法的特点，并提出了联邦学习（FL）辅助的深度强化学习（DRL）框架，该框架通过本地和全球协作共同优化部署策略。案例研究验证了该方法在提高网络部署自动化以及确保数据管理、资源分配和其他任务中的相关KPI方面的有效性。最后，我们讨论了人工智能在部署 6G SAGIN 设置时将面临的重大挑战。