End-to-end network slicing mapping for converged optical-wireless access networks using deep reinforcement learning
LI Ruoyu
WANG Yunwu
GU Jiahua
ZHU Min
Abstract：In converged optical-wireless access networks,optical wavelengths are responsible for carrying wireless data,and their transmission rate in turn affects the allocation of computing and optical bandwidth resources.However,the independent scheduling for the optical and wireless network sides tends to be in-flexible,resulting in inefficient cooperation and utilization of optical and wireless resources.In this paper,we investigate the end-to-end(E2E)optical-wireless network slicing mapping problem in converged opti-cal-wireless access networks.To combine user requirements in the wireless side and radio access network(RAN)slicing scheduling in the optical side,we formulate an E2E network slicing mapping model and pro-pose a deep reinforcement learning(DRL)method.To enhance the decision-making process of DRL a-gent,a slicing request decomposition scheme is proposed,in which each slicing request is divided into two sub-requests.We evaluate the effectiveness of the proposed algorithm through simulations on large-scale 33-node networks.The results demonstrate the superiority of our proposed DRL over the heuristic algo-rithm,achieving an 34.4％reduction in large-scale networks.
Keywords：end-to-end optical-wireless network slicingradio access network(RAN)slicingdeep rein-forcement learning
Publication Date：2025-10-25
Online Publishing Date：2026-08-28(First online date of this platform, not the publication date of the document)
Pages：10( 669-678 )