Joint UAV Trajectory Planning and LEO-Sat Selection in SAGIN
In this paper, the problem of joint unmanned aerial vehicle (UAV) trajectory planning and low-orbit satellites (LEO-Sats) selection in space-air-ground integrated networks (SAGIN) will be investigated. This problem is of utmost importance when SAGIN is exploited for post-disaster relief services, wh...
Saved in:
Published in | IEEE open journal of the Communications Society Vol. 5; pp. 1624 - 1638 |
---|---|
Main Authors | , , |
Format | Journal Article |
Language | English |
Published |
New York
IEEE
2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
Subjects | |
Online Access | Get full text |
ISSN | 2644-125X 2644-125X |
DOI | 10.1109/OJCOMS.2024.3372551 |
Cover
Summary: | In this paper, the problem of joint unmanned aerial vehicle (UAV) trajectory planning and low-orbit satellites (LEO-Sats) selection in space-air-ground integrated networks (SAGIN) will be investigated. This problem is of utmost importance when SAGIN is exploited for post-disaster relief services, where ground base stations (GBSs) within the post-disaster area are completely damaged or malfunctioned. In this scenario, UAV will provide wireless connectivity for the victims, while LEO-Sats will relay the UAV data to the nearest survival GBS. UAV trajectory should be optimized to maximize the collected data from the victims subject to its limited battery capacity, while it should jointly select the best LEO-Sat in each visited location within its optimized trajectory. The selected LEO-Sat should maximize UAV's achievable data rate while maintaining a long remaining visible time to minimize frequent LEO-Sats' handovers. In this paper, an online learning approach using multi-armed bandit (MAB) models will be proposed to address this highly dynamic problem. As LEO-Sat selection should be performed after UAV arrives at a dedicated location in its optimized trajectory, the problem is divided into two MAB stages. In the first stage, the battery constraint UAV trajectory optimization is modeled as budget constraint MAB (BC-MAB) game using BC-upper confidence bound (BC-UCB) algorithm. In the second stage, LEO-Sat selection in each visited location is modeled as contextual MAB with variable arms (CMAB-VA) game using LinUCB-VA algorithm. Numerical analysis confirms the superior performance of the proposed approach over candidate benchmarks. |
---|---|
Bibliography: | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
ISSN: | 2644-125X 2644-125X |
DOI: | 10.1109/OJCOMS.2024.3372551 |