Graduate School and Research Center in Digital Sciences

An efficient and lightweight load forecasting for proactive scaling in 5G mobile networks

Alawe, Imad; Hadjadj-Aoul, Yassine; Ksentini, Adlen; Bertin, Philippe; Viho, Cesar; Darche, Davy

CSCN 2018, IEEE Conference on Standards for Communications and Networking, 29-31 October 2018, Paris, France

The number of connected devices is increasing with the emergence of new services and trends. This phenomenon is leading to a traffic growth over both the control and the data planes of the mobile core network. It is expected that the traffic will increase more and more with the installation of the new generation of mobile networking (5G) as it offers more services that are intended to be connected over the same network, in addition to the legacy ones. Therefore, the 3GPP group has rethought the architecture of the New Generation Core (NGC) by defining its components as Virtualized Network Functions (VNF). However, scalability techniques should be envisioned in order to answer the needs, in term of resource provisioning, without degrading the Quality Of Service (QoS) already offered by hardware based core networks. Neural networks, and in particular deep learning, having shown their effectiveness in predicting time series, could be good candidates for predicting traffic evolution. In this paper, we propose a novel solution to generalize neural networks while accelerating the learning process by using K-means clustering, and a Monte-Carlo method. We benchmarked multiple types of deep neural networks using real operator's data in order to compare their efficiency in forecasting the upcoming network load for dynamic and proactive resources' provisioning. The proposed solution allows obtaining very good predictions of the traffic evolution while reducing by 50% the time needed for the learning phase.

Document Doi Hal Bibtex

Title:An efficient and lightweight load forecasting for proactive scaling in 5G mobile networks
Keywords:5G, Scaling, Mobile Core Network, Prediction, Neural Network, Training
Department:Communication systems
Eurecom ref:5735
Copyright: © 2018 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Bibtex: @inproceedings{EURECOM+5735, doi = {}, year = {2018}, title = {{A}n efficient and lightweight load forecasting for proactive scaling in 5{G} mobile networks}, author = {{A}lawe, {I}mad and {H}adjadj-{A}oul, {Y}assine and {K}sentini, {A}dlen and {B}ertin, {P}hilippe and {V}iho, {C}esar and {D}arche, {D}avy}, booktitle = {{CSCN} 2018, {IEEE} {C}onference on {S}tandards for {C}ommunications and {N}etworking, 29-31 {O}ctober 2018, {P}aris, {F}rance}, address = {{P}aris, {FRANCE}}, month = {10}, url = {} }
See also: