A Survey on Reinforcement Learning Routing Algorithm Research in Delay/Disruption Tolerant Network

Citations

SCOPUS

1

초록

Delay/Disruption Tolerant Networks (DTNs) are designed to operate in highly challenging environments such as deep-space communication, underwater networks, and opportunistic IoT scenarios, where intermittent connectivity and long delays make traditional TCP/IP-based routing infeasible. Reinforcement learning (RL) has emerged as a promising approach to address these challenges by learning adaptive routing strategies aligned with the store-and-forward paradigm of DTNs. Early studies applied Q-learning to optimize next-hop selection and energy efficiency, while recent works have incorporated deep reinforcement learning (DRL) to handle large state spaces and dynamic environments. This survey categorizes RL-based DTN routing schemes into Q-learning approaches, deep RL methods, and hybrid models. For each work, we summarize the applied RL algorithm, network scenario, evaluation metrics, and key contributions. We also highlight how RL has been extended to address security, congestion control, and mobility prediction in DTN contexts. Finally, we identify future directions for advancing RL-based DTN routing.

키워드

6G; Deep RL; Delay/Disruption Tolerant Networks; Q-learning; Reinforcement Learning
제목
A Survey on Reinforcement Learning Routing Algorithm Research in Delay/Disruption Tolerant Network
저자
Kim, Gahyun; Song, Chihyun; Cho, Sungrae
DOI
10.1109/ICTC66702.2025.11388522
발행일
2025
유형
Conference Paper
저널명
International Conference on ICT Convergence
페이지
247 ~ 251