Routing for Space-Air-Ground Integrated Network With GAN-Powered Deep Reinforcement Learning
Qi Guo, Fengxiao Tang, Nei Kato
Abstract
Due to the surge in the development of new applications and services requires high-quality user experiences, characterized by high data transmission rates, low latency, and seamless network connectivity, the space-air-ground integrated network (SAGIN) that combines satellite networks, aerial networks, and terrestrial networks, offering ubiquitous global network services to ground users and enhancing connectivity for a wide range of wireless applications, is rising as the promising architecture for next-generation wireless networks. However, the load-balancing data transmission efficiency in SAGIN remains limited due to the dynamic network topology, long-distance communication links, inefficient real-time network information collection. To address these issues, we construct a free-space optical/radio frequency space-air-ground integrated network that aims to enable large-scale data transmission. Furthermore, we propose a generative adversarial network (GAN)-powered deep reinforcement learning routing strategy to execute dynamic routing in SAGIN while ensuring network load-balancing. The simulation results show that the proposal achieves significant network performance compared with baseline methods.