.png){kind=icon}
.png){kind=icon}
Return
Joint Trajectory and RIS Phase Shift Design for RIS-Assisted SAGIN: A FL-MADRL Approach
.png){kind=icon}
OA
AI
W
R
H
M
K
B
DOI：10.1109/OJVT.2026.3668772.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Wireless systems must deliver ubiquitous QoS for massive IoT, but in urban areas, the blockages severely degrade satellite links. We consider a RIS-enabled space–air–ground integrated network (SAGIN) where a moving LEO satellite provides coverage and multiple drone base stations (DBSs) act as mobile relays to restore connectivity for ground users. Some Reflecting Intelligent Surfaces (RISs) panels, mounted on buildings, create controllable reflect paths to overcome non-line-of-sight conditions. A key challenge is that DBS's 3D trajectory simultaneously affects the satellite–DBS and the DBS-RIS–user access links, while also determining their energy consumption. We formulate a joint optimization problem to maximize the communication coverage, system throughput, and fairness under mobility, energy, and connectivity constraints. To address the issue, we develop a Federated Multi-Agent Deep Reinforcement Learning (FL-MADRL) method that coordinates multiple agents: each agent learns online from local experience, while periodic federated aggregation shares policies without exchanging raw data. Simulations in an urban scenario demonstrate that the proposed approach maintains uninterrupted service in terms of coverage, throughput, and fairness while reducing DBS energy consumption compared to conventional convex approaches and non-federated baselines, indicating its suitability for RIS-assisted SAGIN systems.
Keywords:
Low earth orbit (LEO)
drone base station (DBS)
reflecting intelligent surfaces (RISs)
federated multi-agent deep reinforcement learning (FL-MADRL)
3D trajectory control
energy consumption
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
I
IF:
4.8
Papers:
493
Citations:
987
Organization
Cited Papers