Impact of Multi-Vendor Transponders Performance on Design Margin in Optical Networks
| Autor: | Konstantinos Christodoulopoulos, Ricardo Martinez, Raul Munoz, Ankush Mahajan, Salvatore Spadaro |
|---|---|
| Přispěvatelé: | Universitat Politècnica de Catalunya. Departament de Teoria del Senyal i Comunicacions, Universitat Politècnica de Catalunya. GCO - Grup de Comunicacions Òptiques |
| Rok vydání: | 2021 |
| Předmět: |
General Computer Science
Vendor Computer science 020209 energy optical networks 02 engineering and technology Reduction (complexity) Signal-to-noise ratio Margin (machine learning) Wavelength-division multiplexing Transponders 0202 electrical engineering electronic engineering information engineering Comunicacions òptiques General Materials Science Design margin quality of transmission (QoT) transponders Optical communications General Engineering Physical layer Design Margin Optical Networks Enginyeria de la telecomunicació [Àrees temàtiques de la UPC] Reliability engineering TK1-9971 Network planning and design NO KEYWORDS Transmission (telecommunications) Quality Of Transmission (Qot) 020201 artificial intelligence & image processing Electrical engineering. Electronics. Nuclear engineering |
| Zdroj: | IEEE Access UPCommons. Portal del coneixement obert de la UPC Universitat Politècnica de Catalunya (UPC) IEEE Access, Vol 9, Pp 118927-118940 (2021) |
| ISSN: | 2169-3536 |
| DOI: | 10.1109/access.2021.3107296 |
| Popis: | For reliable and efficient network planning and operation, accurate estimation of Quality of Transmission (QoT) is necessary. In optical networks, a physical layer model (PLM) is typically used as a QoT estimation tool (Qtool) including a design margin to account for modeling and parameter inaccuracies, to ensure acceptable performance. Such margin also covers the performance variations of the transponders (TPs) which are relatively low in a single vendor environment. However, for disaggregated networks that utilize TPs from multiple vendors, such as partial disaggregated networks with open line system (OLS), this traditional approach limits the Qtool estimation accuracy. Although higher TP performance variations can be covered with an additional margin, this approach would reduce the efficiency and consume the benefits of disaggregation. Therefore, we propose PLM extensions that capture the performance variations of multi- vendor TPs. In particular, we propose four TP vendor dependent performance factors and we also devise a Machine Learning (ML) scheme to learn these performance factors in offline and online network planning scenarios. The proposed extended PLM and ML training scheme are evaluated through realistic simulations. Results show a design margin reduction of greater than 1 dB for new connection requests in a disaggregated network with TPs from four vendors. On top of this, the results also show a ~0.5 dB additional Signal to Noise Ratio (SNR) saving for new connection requests by proper selection of the TPs. This work is a part of the Future Optical Networks for Innovation, Research and Experimentation (ONFIRE) Project (Project ID-765275). |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|