Machine Learning Classification and Regression Approaches for Optical Network Traffic Prediction

Abstract

Rapid growth of network traffic causes the need for the development of new network technologies. Artificial intelligence provides suitable tools to improve currently used network optimization methods. In this paper, we propose a procedure for network traffic prediction. Based on optical networks’ (and other network technologies) characteristics, we focus on the prediction of fixed bitrate levels called traffic levels. We develop and evaluate two approaches based on different supervised machine learning (ML) methods—classification and regression. We examine four different ML models with various selected features. The tested datasets are based on real traffic patterns provided by the Seattle Internet Exchange Point (SIX). Obtained results are analyzed using a new quality metric, which allows researchers to find the best forecasting algorithm in terms of network resources usage and operational costs. Our research shows that regression provides better results than classification in case of all analyzed datasets. Additionally, the final choice of the most appropriate ML algorithm and model should depend on the network operator expectations.

Application of machine learning in automatic fault detection for passive optical network

“…The downlink speed of GPON is twice than EPON, which can well support broadband full service access and has richer OAM capabilities. Therefore, it is widely adopted in future networks [7] .…”

Threshold-Based Routing-Topology Co-Design for Optical Data Center

“…RDC [33] and Hedera [33] predict traffic demands by the max-min fair share rate of all flows within a network. Other works [34], [35], [36], [37], [38], [39], [40], [41] use machine learning models, Markov chain probability models, etc., to predict traffic patterns. However, none of the above methods can accurately predict traffic.…”

Long-Term Traffic Forecasting in Optical Networks Using Machine Learning

“…To the best of our knowledge, long-term traffic forecasting has not been addressed in the literature in the context of prediction of traffic level. Additionally, short-term traffic forecasting using traffic levels was described only in articles [14], [25] and [26]. To fill the research gap, this work introduces, formulates, and examines the long-term forecasting problem as a prediction of fixed traffic levels.…”

“…• We propose an evaluation metric suitable for the considered problem. It is based on metric described in [14] and allows to evaluate tested algorithms in terms of underpredictions and overpredictions, which can be significant for network operators. Its main characteristic is flexibility.…”

“…In order to estimate the performance of algorithms, thus final traffic level forecasting quality, in this work, the metric called Traffic Level Prediction Quality (TLPQ) is defined. It is an extended version of the metric, which we presented in[14]. Compared it to its previous version, it considers more under and overpredictions variants.…”