Abstract:
We propose a trajectory-planning method that utilizes deep Q-network-based multi-agent reinforcement learning architecture for multiple fixed-wing base stations. In our s...Show MoreMetadata
Abstract:
We propose a trajectory-planning method that utilizes deep Q-network-based multi-agent reinforcement learning architecture for multiple fixed-wing base stations. In our system model, the trajectory-planning problem is proposed to maximize the coverage area in the wireless sensor network. For multi-agent learning, we propose an architecture where a single Q-network is shared across all base stations. The result shows that our model can stably converge and almost achieve the performance of a single base station, which is not interrupted by other base stations. Moreover, we propose a penalty function to regulate the base stations when the trajectories are undesirable. The result shows that the penalty function can stabilize the network.
Published in: 2022 13th International Conference on Information and Communication Technology Convergence (ICTC)
Date of Conference: 19-21 October 2022
Date Added to IEEE Xplore: 25 November 2022
ISBN Information: