AI-Driven Seamless and Massive Access in Space-Air-Ground Integrated Networks
Zhi Lin, Zimo Feng, Kefeng Guo, Ali Nauman, Dusit Niyato, Jiangzhou Wang
Abstract
The global ubiquitous coverage vision of the sixth-generation (6G) mobile communication system marks a significant leap beyond its predecessors by combining terrestrial networks (TN) with non-terrestrial networks (NTN) into a unified framework known as space-air-ground integrated networks (SAGIN). With the proliferation of IoT devices and reliable connectivity demands in diverse environments, seamless and massive access has become one of the key indicators of future networks which still suffer from frequently changed network topology, limited spectrum resources, dynamic and unstable links, and constrained energy supply. Therefore, it is critical to study how to leverage the potential of artificial intelligence (AI) to achieve highly efficient multiple access in SAGIN. To address these challenges, we first introduce the structure and key issues of SAGIN. Then, recent advanced multiple access techniques are thoroughly analyzed and compared. In particular, we explore a practical AI case for joint optimization of multidimensional resources in SAGIN and further analyze why the deep reinforcement learning (DRL) method can achieve efficient resource management and performance enhancement compared to traditional optimization methods based on their principles and simulation results. Finally, several future research directions are outlined to promote the development of AI-driven massive access in SAGIN.