Reinforcement Learning for Power Management in Low-margin Optical Networks
Paper in proceeding, 2024
We explore the Q-learning and Proximal Policy Optimization (PPO) algorithms for solving the Routing, Wavelength, and Power Allocation (RWPA) problem. The results indicate that reinforcement learning can significantly optimize launch power in low-margin optical networks, with Q-learning achieving better results than PPO, delivering improvement in GSNR of over 175 % and over 48 % in small and large networks, respectively.
optical networks
power management
q-learning
Author
Sze Ka Tse
Khoury College of Computer Sciences
Xiaoyang Zhao
Khoury College of Computer Sciences
Anita Chan
Khoury College of Computer Sciences
Di Tang
Khoury College of Computer Sciences
Aanchan Mohan
Khoury College of Computer Sciences
Carlos Natalino Da Silva
Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks
Michal Aibin
Khoury College of Computer Sciences
British Columbia Institute of Technology
International Conference on Transparent Optical Networks
21627339 (ISSN)
9798350377309 (ISBN)
24th International Conference on Transparent Optical Networks, ICTON 2024
Bari, Italy,
Bari, Italy,
Subject Categories
Telecommunications
DOI
10.1109/ICTON62926.2024.10647982