Presentation Information

[WP-A-3]Dynamic RWA for Programmable Filterless Optical Networks Based on Deep Reinforcement Learning

zhaoyang liu1, taoning zheng1, tingyi yao1, xiangyu ge1, yi fang1, ○bitao pan1 (1.Beijing University of Posts and Telecommunications)

Keywords:

Artificial intelligence and machine learning for optical network design,control,and management

We propose a graph convolutional network enhanced reinforcement learning framework for dynamic resource allocation in programmable filterless optical networks. It realized 9.2% and 19% blocking rate reductions compared to two benchmarks with lower wavelength waste.