AI-Enabled Deployment Automation for 6G Space-Air-Ground Integrated Networks: Challenges, Design, and Outlook

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.