Regressor Relearning Architecture Adapting to Traffic Trend Changes in NFV Platforms

Network function virtualization (NFV) enables network operators to flexibly provide diverse virtualized functions for services such as Internet of things (IoT) and mobile applications. NFV platforms are required to offer stable and guaranteed quality-of-service (QoS)even during dynamically changing resource demands and traffic volumes. To meet the QoS requirements against time-varying network environments, infrastructure providers must dynamically adjust the amount of computational resources, such as CPU, assigned to virtual network functions (VNFs). To provide agile resource control and adaptiveness, predicting the virtual server load via machine learning technologies is an effective approach for proactive control. In this paper, we propose a traffic prediction framework based on ensemble learning, comprising weak regressors trained by ML models, such as recurrent neural networks (RNNs), random forest, and elastic net. It was observed that the prediction error tends to worsen with time because the gap of trends between the past and future traffics becomes wider. Therefore, to reduce the prediction errors, we propose an adjustment mechanism for regressors based on forgetting and dynamic ensemble. The evaluation result with real traffic data verified that the resource adjustment scheme based on the proposed traffic prediction framework keeps the frequency of over and under provisioning low, which is lesser by 45% in comparison to RNNs and autoregressive moving average (ARMA).

I. Introduction

Network function virtualization (NFV) techniques enable us to implement the softwarized functions of network equipment, known as virtual network functions (VNFs), on generic servers [1]. A service function (SF) chain contains a series of VNFs, such as load balancer, firewall, intrusion detection system (IDS), and contents server [2]. If the SF chains are created manually according to a legacy VNF construction procedure, it may take approximately two weeks for the successful construction of an SFC, from receiving a construction request to providing the service to the customer [3]. To continue satisfying the diverse quality-of-service (QoS) requirements despite time-varying network user behaviors and traffic volumes, the automation of SFC management is essential to shorten the network configuration/reconfiguration time and quickly rolling out a new service. The automation of the deployment and periodical adjustments of the computational resources for each network function (NF), in particular, would help in realizing efficient and stable service provisioning to customers in time-varying network conditions (e.g., resource utilization and failure occurrences).