5G is expected to provide network connectivity to not only classical devices (i.e. tablets, smartphones, etc) but also to the Internet Of Things (IOT), which will drastically increase the traffic load carried over the network. 5G will mainly rely on Network Function Virtualization (NFV) and Software Defined Network (SDN) to build flexible and on-demand instances of functional networking entities, via Virtual Network Functions (VNF). Indeed, 3GPP is devising a new architecture for the core network, which replaces point to point interfaces used in 3G and 4G, by a producer/consumer-based communication among 5G core network functions, facilitating deployment over a virtual infrastructure. One big advantage of using VNF, is the possibility of dynamically scaling, depending on traffic load (i.e. instantiate new resources to VNF when the traffic load increases, and reduce the number of resources when the traffic load decreases). In this paper, we propose a novel mechanism to scale 5G core network resources by anticipating traffic load changes through forecasting via Machine Learning (ML) techniques. The traffic load forecast is achieved by using and training a Neural Network on a real dataset of traffic arrival in a mobile network. Two techniques were used and compared: (i) Recurrent Neural Network (RNN), more specifically Long Short Term Memory Cell (LSTM); and (ii) Deep Neural Network (DNN). Simulation results showed that the forecast-based scalability mechanism outperforms the threshold-based solutions, in terms of latency to react to traffic change, and delay to have new resources ready to be used by the VNF to react to traffic increase.

Abstract-One of the requirement of 5G is to support massive number of connected devices, considering many use-cases such as IoT and massive Machine Type Communication (MTC). While this represents an interesting opportunity for operators to grow their business, it will need new mechanisms to scale and manage the envisioned high number of devices and their generated traffic. Particularity, the signaling traffic, which will overload the 5G core Network Function (NF) in charge of authentication and mobility, namely Access and Mobility Management Function (AMF). The objective of this paper is to provide an algorithm based on Control Theory allowing: (i) to equilibrate the load on the AMF instances in order to maintain an optimal response time with limited computing latency; (ii) to scale out or in the AMF instance (using NFV techniques) depending on the network load to save energy and avoid wasting resources. Obtained results via computer system indicate the superiority of our algorithm in ensuring fair load balancing while scaling dynamically with the traffic load.

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 should be in a manner that all the needed resources will be deployed and active when the load increases. In the same way, it will scalein 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 compared to a thresholdbased scaling technique.

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 proposed a novel solution to generalize neural networks while accelerating the learning process by using K-mean 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 predicting the upcoming network load for dynamic and proactive resource provisioning. The proposed solution allow obtaining very good predictions of the traffic evolution while reducing by 50% the time needed for the learning phase.

-Software Defined Networking (SDN) is one of the key enablers for evolving mobile network architecture towards 5G. SDN involves the separation of control and data plane functions, which leads in the context of 5G, to consider the separation of the control and data plane functions of the different gateways of the Evolved Packet Core (EPC), namely Serving and Packet data Gateways (S and P-GW). Indeed, the envisioned solutions propose to separate the S/P-GW into two entities: the S/P-GW-C, which integrates the control plane functions; and the S/P-GW-U that handles the User Equipment (UE) data plane traffic. There are two major approaches to create and update user plane forwarding rules for such a partition: (i) considering an SDN controller for the S/P-GW-C (SDNEPC) or (ii) using a direct specific interface to control the S/P-GW-U (enhancedEPC). In this paper, we evaluate, using a testbed, those two visions against the classical virtual EPC (vEPC), where all the elements of the EPC are virtualized. Besides evaluating the capacity of the vEPC to manage and scale to UE requests, we compare the performances of the solutions in terms of the time needed to create the user data plane. The obtained results allow drawing very interesting remarks, which may help to dimension the vEPC's components as well as to improve the S/P-GW-U management procedure.