面向多域网络的智能化自动网络控制系统

Intelligent Network Automation System for the Control of Multi-domain Networks

提出了一种新的面向大规模网络的深度学习(DL)辅助控制平面(CP)系统DeepMDR,用于实现多域分组网络中可伸缩的、独立于协议的路径计算。我们基于ONOS平台开发了DeepMDR,并扩展该系统以支持协议无感知转发(POF),简化多域CP操作的层次结构,设计并实现了一个DL模型进行域内快速高质量的路径计算。文章通过数值仿真和实验验证了该系统的可行性和有效性。

In this paper,we propose a new deep learning(DL)assisted control plane(CP)system for large scale network,called DeepMDR,which realizes scalable and protocol-independent path computation in multi-domain packet networks.We develop DeepMDR based on ONOS,extend it to support protocol-oblivious forwarding(POF)and facilitate the hierarchical architecture for multi-domain CP operations,and design and implement a DL model to achieve fast and high-quality path computation in each domain.Both numerical simulations and experiments are performed to verify the effectiveness and feasibility of our proposal.