TY - GEN
T1 - Satellite Staring Beam Scheduling Strategy Based on Multi-agent Reinforcement Learning
AU - Zhu, Hongtao
AU - Wang, Zhenyong
AU - Li, Dezhi
AU - Guo, Qing
N1 - Publisher Copyright:
© 2022, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.
PY - 2022
Y1 - 2022
N2 - Low Earth Orbit (LEO) satellites are an important part of Space-Air-Ground Integrated Networks (SAGIN), which play an irreplaceable role in providing global communication and emergency communication. With the development of phased array technology, many satellites begin to try to use staring beam technology, which can make the beam serve a hot spot on the ground as long as possible by adjusting its phased array parameters, so as to reduce the impact of fast switching on the service performance of LEO satellites. In the satellite service time, how to balance the load of each satellite and meet the communication needs of hot spots is an important problem to be considered. Excellent beam allocation strategy can reduce the network handover rate and signaling overhead. In this paper, the satellite staring beam scheduling problem is transformed into a two-dimensional model, and we propose a novel satellite beam scheduling strategy based on multi-agent reinforcement learning that aims to maximize system performance. Each satellite is regarded as an individual agent, and the decision is to provide communication beam for the current hot spot area. Compared with the beam allocation algorithm based on KM, simulation results show that the proposed strategy can effectively reduce the handoff rate of hot spots when the coverage is satisfied.
AB - Low Earth Orbit (LEO) satellites are an important part of Space-Air-Ground Integrated Networks (SAGIN), which play an irreplaceable role in providing global communication and emergency communication. With the development of phased array technology, many satellites begin to try to use staring beam technology, which can make the beam serve a hot spot on the ground as long as possible by adjusting its phased array parameters, so as to reduce the impact of fast switching on the service performance of LEO satellites. In the satellite service time, how to balance the load of each satellite and meet the communication needs of hot spots is an important problem to be considered. Excellent beam allocation strategy can reduce the network handover rate and signaling overhead. In this paper, the satellite staring beam scheduling problem is transformed into a two-dimensional model, and we propose a novel satellite beam scheduling strategy based on multi-agent reinforcement learning that aims to maximize system performance. Each satellite is regarded as an individual agent, and the decision is to provide communication beam for the current hot spot area. Compared with the beam allocation algorithm based on KM, simulation results show that the proposed strategy can effectively reduce the handoff rate of hot spots when the coverage is satisfied.
KW - Low orbit satellite
KW - Multi-agent reinforcement learning
KW - Staring beam scheduling
UR - https://www.scopus.com/pages/publications/85124125376
U2 - 10.1007/978-3-030-93398-2_3
DO - 10.1007/978-3-030-93398-2_3
M3 - 会议稿件
AN - SCOPUS:85124125376
SN - 9783030933975
T3 - Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST
SP - 23
EP - 34
BT - Wireless and Satellite Systems - 12th EAI International Conference, WiSATS 2021, Proceedings
A2 - Guo, Qing
A2 - Meng, Weixiao
A2 - Jia, Min
A2 - Wang, Xue
PB - Springer Science and Business Media Deutschland GmbH
T2 - 12th International Conference on Wireless and Satellite Services, WiSATS 2021
Y2 - 31 July 2021 through 2 August 2021
ER -