Virtual Network Topology Adaptability Based on Data Analytics for Traffic Prediction

Abstract

Abstract-The introduction of new services requiring large and dynamic bitrate connectivity can cause changes in the direction of the traffic in metro and even core network segments along the day. This leads to large overprovisioning in statically managed virtual network topologies (VNT), designed to cope with the traffic forecast. To reduce expenses while ensuring the required grade of service, in this paper we propose the VNT reconfiguration approach based on data analytics for traffic prediction (VENTURE); it regularly reconfigures the VNT based on predicted traffic thus, adapting the topology to both, the current and the predicted traffic volume and direction. A machine learning algorithm based on artificial neural network (ANN) is used to provide robust and adaptive traffic models. The reconfiguration problem that takes as input the traffic prediction is modelled mathematically and a heuristic is proposed to solve it in practical times. To support VENTURE, we propose an architecture that allows collecting and storing data from monitoring at the routers and that is used to train predictive models for every origin-destination pair. Exhaustive simulation results of the algorithm together with the experimental assessment of the proposed architecture are finally presented.

Dynamic Core VNT Adaptability Based on Predictive Metro-Flow Traffic Models

“…Once these models are available, the VENTURE problem is executed at regular intervals (e.g. hourly) to adapt the VNT to the future traffic conditions [5].…”

“…Limiting the model evaluation to the mean and the variance as presented in Table II discards other interesting estimations such as the maximum bitrate, important to re-optimize the VNT [5]. Although this estimation is not directly provided by the algorithm, we can obtain it in a later stage by applying results from probability theory involving μ and σ 2 .…”

“…To support and automate VNT adaptability, authors in [5] proposed a data analytics -enabled network manager architecture to store OD traffic monitoring data featuring prediction and detection of traffic anomalies [6] and network reconfiguration in response [7]. Predictive traffic models can be obtained by applying machine learning to monitored traffic data; e.g., artificial neural networks (ANN) were considered in [5].…”

“…Predictive traffic models can be obtained by applying machine learning to monitored traffic data; e.g., artificial neural networks (ANN) were considered in [5]. Those predictive traffic models can be used in the process of VNT reconfiguration, where the VNT can be proactively adapted for the predicted near future traffic conditions; this improves resource utilization and reduces overprovisioning.…”

“…Those predictive traffic models can be used in the process of VNT reconfiguration, where the VNT can be proactively adapted for the predicted near future traffic conditions; this improves resource utilization and reduces overprovisioning. In particular, the VENTURE problem was proposed in [5] to periodically adapt the VNT to current and predicted traffic.…”