Smart scaling of the 5G core network: an RNN-based approach

The upcoming mobile core network, which will be based on Virtual Network Functions (VNF), will face an increase of data traffic on both data and control planes. This is due

to the increase of the number of connected devices and the newly 5G supported-services like IoT, Connected Health Care etc. Therefore dynamic and accurate 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. Although provisioning new resources is easier as it is a matter of software deployment, the strategy to use (when to scale and how much

to scale) remains complex. In this paper we propose scaling techniques based on neural networks to forecast the upcoming load. Hence scheduling the resource provisioning in a manner that all the needed resources will be deployed and active when the load increases. In the same way, it will scale-in the unneeded resources when the traffic load decreases. The proposal is tested via discrete event simulations using a traffic load dataset provided

by a Network Operator. The results show clearly the robustness of our proposal comparing to a threshold-based scaling technique.