Reinforcement Learning-Based Energy-Efficient Data Access for Airborne Users in Civil Aircrafts-Enabled SAGIN

Qian Chen, Weixiao Meng, Shuai Han, Cheng Li, Hsiao Hwa Chen

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)


Airborne users are always dreaming of enjoying a good Internet access experience while in the air. However, due to long propagation delay and limited network coverage, the existing data communication methods utilized in space and ground communications not only fail to ensure the quality-of-service (QoS) of airborne users, but also incur significant energy consumption to process content requests. In this paper, we introduce the aeronautical ad hoc network (AANET) as a new method of network access and design an energy-efficient data access scheme in civil aircrafts-enabled space-air-ground integrated networks (CAE-SAGIN). In order to minimize the energy consumption, we propose a service selection scheme based on reinforcement learning and formulate a joint optimization problem of resource allocation and request distribution. Leveraged by the Lyapunov optimization method, the optimization problem can be solved by the proposed joint optimization algorithm. Extensive simulations are conducted to confirm the stability of the CAE-SAGIN, and demonstrate that the proposed data access scheme can effectively reduce both the energy consumption and the processing delay. Moreover, the advantages of using AANET are becoming more obvious when higher data rate is required.

Original languageEnglish
Article number9361631
Pages (from-to)934-949
Number of pages16
JournalIEEE Transactions on Green Communications and Networking
Issue number2
Publication statusPublished - 2021 Jun

All Science Journal Classification (ASJC) codes

  • Renewable Energy, Sustainability and the Environment
  • Computer Networks and Communications


Dive into the research topics of 'Reinforcement Learning-Based Energy-Efficient Data Access for Airborne Users in Civil Aircrafts-Enabled SAGIN'. Together they form a unique fingerprint.

Cite this