Tsukada Lab. Computer Network Group

Adaptive Entropy-Regularized Reinforcement Learning for Routing and Slot Allocation in MTNs

Lin Cong, Ying Wang, Peng Yu, Xuesong Qiu, Shaoyong Guo, Manabu Tsukada, Long Qu, Jice Liu

IEEE/IFIP Network Operations and Management Symposium (NOMS 2026), Rome, Italy, 2026年5月18日

3行要約

  • EFP-RSAは都市圏の基幹網で、経路と伝送スロットの割り当てを学習する。
  • 方策を経路選択とスロット選択に分けて連携させ、学習過程でエントロピー正則化を調整する。
  • シミュレーションでは比較手法より接続拒否が減り、スロットの利用率が高まった。

アブストラクト(原文)

As emerging services such as AI datacenter interconnection, immersive AR/VR, and autonomous driving demand greater transport flexibility and resource efficiency, Metro Transport Networks (MTNs) have become the critical infrastructure for next-generation metro backbones providing flexible and efficient resource orchestration. Nevertheless, the Routing and Slot Allocation (RSA) problem in MTNs remains challenging due to combinatorial complexity, high-dimensional coupling in the action space, and resource fragmentation. We propose EFP-RSA, a novel entropy-regularized reinforcement learning framework that addresses the RSA problem through factorized policy. It factorizes the joint optimization into coordinated path selection and slot assignment, alongside entropy regularization annealing to balance exploration with convergence. Compared with baselines, EFP-RSA reduces blocking probability by 35.3%–60.7% and increases slot utilization by 19.4%–58.6%, while maintaining low load variance and high solution efficiency, demonstrating strong generalization and scalability in complex MTN scenarios.

引用方法

L. Cong, Y. Wang, P. Yu, X. Qiu, S. Guo, M. Tsukada, L. Qu, and J. Liu, “Adaptive Entropy-Regularized Reinforcement Learning for Routing and Slot Allocation in MTNs,” in IEEE/IFIP Network Operations and Management Symposium (NOMS 2026), Rome, Italy, 2026, doi: 10.1109/NOMS69089.2026.11668250.

BibTeX

@inproceedings{Cong2026,
  title         = {Adaptive Entropy-Regularized Reinforcement Learning for Routing and Slot Allocation in MTNs},
  author        = {Lin Cong and Ying Wang and Peng Yu and Xuesong Qiu and Shaoyong Guo and Manabu Tsukada and Long Qu and Jice Liu},
  doi           = {10.1109/NOMS69089.2026.11668250},
  year          = {2026},
  date          = {2026-05-18},
  urldate       = {2026-05-18},
  booktitle     = {IEEE/IFIP Network Operations and Management Symposium (NOMS 2026)},
  address       = {Rome, Italy},
  abstract      = {As emerging services such as AI datacenter interconnection, immersive AR/VR, and autonomous driving demand greater transport flexibility and resource efficiency, Metro Transport Networks (MTNs) have become the critical infrastructure for next-generation metro backbones providing flexible and efficient resource orchestration. Nevertheless, the Routing and Slot Allocation (RSA) problem in MTNs remains challenging due to combinatorial complexity, high-dimensional coupling in the action space, and resource fragmentation. We propose EFP-RSA, a novel entropy-regularized reinforcement learning framework that addresses the RSA problem through factorized policy. It factorizes the joint optimization into coordinated path selection and slot assignment, alongside entropy regularization annealing to balance exploration with convergence. Compared with baselines, EFP-RSA reduces blocking probability by 35.3%–60.7% and increases slot utilization by 19.4%–58.6%, while maintaining low load variance and high solution efficiency, demonstrating strong generalization and scalability in complex MTN scenarios.},
  keywords      = {Other},
  pubstate      = {published},
  tppubtype     = {inproceedings}
}