TY  - JOUR
T1  - The Impact of Dataset on Offline Reinforcement Learning of Multiple UAVs for Flying Ad-hoc Network Formation
AU  - Lee, Dongsu 
AU  - Kwon, Minhae 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2024
DA  - 2024/1/1
DO  - 10.7840/kics.2024.49.7.1002
KW  - Artificial Intelligence
KW  - Multi-agent Reinforcement Learning
KW  - Offline Reinforcement Learning
KW  - Ad-hoc Network
AB  - A Flying Ad-hoc Network (FANET) is a network of airborne mobile nodes that can communicate independently, and can be utilized as a fallback network during a crisis, such as a disaster or war, when existing infrastructures are damaged. Several Unmanned Aerial Vehicles (UAVs) with communication and routing capabilities can travel to areas where communication is not possible and establish a FANET. In this study, we consider a scenario where multiple UAVs trained through offline reinforcement learning establish a FANET without a centralized control. We conducted experiments to compare the performance of multi-agent’s network construction by datasets and offline reinforcement learning algorithms, and analyzed the learning aspects that depend on the features of datasets and algorithms.