Summary
2021
Session Number:PS2
Session:
Number:PS2-14
Applying RouteNet and LSTM to Achieve Network Automation: An Intent-Based Networking Approach
Talha Ahmed Khan, Khizar Abbas, Javier Jose Diaz Rivera, Afaq Muhammad, Wang-Cheol Song,
pp.254-257
Publication Date:2021/9/8
Online ISSN:2188-5079
Summary:
The expansion of infrastructure and services in the 5th generation networks resulted in complex configuration management throughout the network lifecycle. To this end, network automation replaces existing traditional manual administrative approaches with software-driven repetitive and reliable applications. Since network expansion is in multiple dimensions, including multi-services, domains, and platforms, it is challenging to resolve such a vast infrastructure through a single automation solution. Hence this paper proposed applying an intent-based solution for achieving automatic orchestration for vastly spreading network services. Intent-based solution not only considers network automation but also performs service assurance throughout the network service lifecycle. The proposed IBN (Intent-Based Networking) solution implements a closed-loop network lifecycle management using a single abstracted software platform. It translates high-level requirements to the infrastructure irrespective of the various underlying platforms and domains, and it includes intelligence-driven monitoring and updates for service assurance. A multi-model machine learning approach is proposed in this work to control the network infrastructure reliably. To this end LSTM (Long Short-Term Memory) algorithm is applied for compute-resource prediction and the Route-Net model for optimized service path routing. The infrastructure includes FlexRAN deployed as the access network controller, OSM (OpenSource MANO) resides at the core, and the KOREN network serves as a high-speed transport network.