Dynamic Traffic Prediction Model Retraining for Autonomous Network Operation

Fatemeh Tabatabaeimehr (1) , Luis Velasco (2) , Marc Ruiz (2) , Hamzeh Khalili (1) , Ramon Aparicio-Pardo (3, 4)

1 CTTC - Centre Tecnològic de Telecomunicacions de Catalunya = Telecommunications Technological Centre of Catalonia
2 UPC - Universitat Politècnica de Catalunya = Université polytechnique de Catalogne [Barcelona]
3 Laboratoire d'Informatique, Signaux, et Systèmes de Sophia-Antipolis (I3S) / Equipe SIGNET
4 COATI - Combinatorics, Optimization and Algorithms for Telecommunications

In general, the availability of an accurate machine learning (ML) model plays a particularly important role in the development of new networking solutions and is one of the main drivers for the development of 5G and beyond networking. Although an option is to update the model once inaccurate data is detected, such approach requires high computational effort, specially once the data history is large. In this paper, we propose an approach that combines a traffic prediction model based on Long Short-Term Memory (LSTM) with an analysis module for dynamic connection capacity allocation. Once the model is generated, retraining can be triggered after inaccuracies are detected by the analysis module. Illustrative numerical results show the benefits from the proposed decision-based retraining approach to reduce the number of retraining rounds while maintaining model accuracy.

Mots clés: autonomous operation metro network traffic prediction model retraining LSTM; autonomous operation; metro network; traffic prediction; model retraining; LSTM

Domaines: Réseaux et télécommunications [cs.NI]; Intelligence artificielle [cs.AI]

Fatemeh Tabatabaeimehr, Luis Velasco, Marc Ruiz, Hamzeh Khalili, Ramon Aparicio-Pardo. Dynamic Traffic Prediction Model Retraining for Autonomous Network Operation. ICTON 2023 - 23rd International Conference on Transparent Optical Networks, Jul 2023, Bucharest, Romania. ⟨10.1109/ICTON59386.2023.10207375⟩. ⟨hal-04135969⟩