Enabling traffic forecasting with cloud-native SDN controller in transport networks
Network bandwidth is a scarce resource that network operators monitor to cope with future traffic demands and plan more transceiver and fibre deployments. The inclusion of Machine Learning permits the usage of traffic forecasting methods to predict future link usage. Typically, traffic analysis is p...
| Autores: | , , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2024 |
| País: | España |
| Institución: | Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| Repositorio: | r-CTTC. Repositorio Institucional Producción Científica del Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| OAI Identifier: | oai:cttc.fundanetsuite.com:p8447 |
| Acceso en línea: | https://cttc.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=8447 https://www.scopus.com/inward/record.uri?eid=2-s2.0-85196311381&doi=10.1016%2fj.comnet.2024.110565&partnerID=40&md5=49bd2239831eeff9541adb0f225242ca |
| Access Level: | acceso abierto |
| Palabra clave: | Forecasting Network architecture Topology Forecasting methods Link usage Network bandwidth Network operator Scarce resources SDN Traffic demands Traffic Forecasting Traffic plan Transport networks Controllers |
| Sumario: | Network bandwidth is a scarce resource that network operators monitor to cope with future traffic demands and plan more transceiver and fibre deployments. The inclusion of Machine Learning permits the usage of traffic forecasting methods to predict future link usage. Typically, traffic analysis is performed offline due to the high computational load and difficulty of obtaining real-time data directly from the underlying network devices. To overcome these limitations, this paper presents and evaluates an architecture for SDN-controlled packetoptical transport networks to allow real-time traffic monitoring in the transport SDN controller. The presented SDN controller is based on a micro-service-based architecture, which facilitates the ease of deployment of the proposed solution. Four forecasting methods are proposed and evaluated against two topologies to select the most precise and the fastest among them.The algorithm random forest seems to be the most accurate forecasting future link usage with 79.98 % and 95.88 % accuracy and a reasonable fast speed when implemented it into two different topologies © 2024 Elsevier B.V. |
|---|