5G is the next telecommunications standards that will enable the sharing of physical infrastructures to provision ultra shortlatency applications, mobile broadband services, Internet of Things, etc. Network slicing is the virtualization technique that is expected to achieve that, as it can allow logical networks to run on top of a common physical infrastructure and ensure service level agreement requirements for different services and applications. In this vein, our paper proposes a novel and complete solution for planning network slices of the LTE EPC, tailored for the enhanced Mobile BroadBand use case. The solution defines a framework which consists of: i) an abstraction of the LTE workload generation process, ii) a compound traffic model, iii) performance models of the whole LTE network, and iv) an algorithm to jointly perform the resource dimensioning and network embedding. Our results show that the aggregated signaling generation is a Poisson process and the data traffic exhibits self-similarity and long-range-dependence features. The proposed performance models for the LTE network rely on these results. We formulate the joint optimization problem of resources dimensioning and embedding of a virtualized EPC and propose a heuristic to solve it. By using simulation tools, we validate the proper operation of our solution.

The fifth generation (5G) and beyond wireless networks are foreseen to operate in a fully automated manner, in order to fulfill the promise of ultra-short latency, meet the exponentially increasing resource requirements, and offer the quality of experience (QoE) expected from end- users. Among the ingredients involved in such environments, network slicing enables the creation of logical networks tailored to support specific application demands (i.e., service level agreement SLA, quality of service QoS, etc.) on top of physical infrastructure. This creates the need for mechanisms that can collect spatiotemporal information on users' service consumption, and identify meaningful insights and patterns, leveraging machinelearning techniques. In this vein, our paper proposes a framework dubbed "SOCL" for the Service Oriented CLustering, analysis and profiling of users (i.e., humans, sensors, etc.) when consuming enhanced Mobile BroadBand (eMBB) applications, internet of things (IoT) services, and unmanned aerial vehicles services (UAVs). SOCL relies mainly on the realistic network simulation framework "network slice planner" (NSP), and two clustering methods namely K-means and hierarchical clustering. The obtained results showcase interesting features, highlighting the benefit of the proposed framework.

The key concept beneath Multi-Access Edge Computing (MECs) is to place cloud resources in closer proximity to end-users, through the installation of small-scale cloud infrastructures at the network edge. In MEC environments, we identify two issues: 1) data about users' activities are not always available, and 2) the available virtual resource planning mechanisms (i.e., algorithms for the placement of Virtual Network Functions - VNFs) are not efficient enough to fulfill the QoS requirements and deployment costs. In this vein, we design a layered framework to define the presence of Mobile BroadBand User Equipments (UEs) and automate the underlying virtual resource placement and management based on the Fuzzy Logic Controller paradigm (FLC). Experimentation results show that our framework, compared to baseline solutions, achieves good performance results; the end-to-end delay is enhanced by 25%, the resource consumption is reduced by 30%, and the environmental impact, reflected by the carbon footprint that depends on the amount of deployed Virtual Machines (VMs), is reduced by 50%.

The network softwarization paradigm, enabled by Network Function Virtualization (NFV), facilitates the automation of management operations and orchestration of future networks, thus reducing their operational expenditures. The envisioned management practices include the introduction of automation in the scaling of network services. This may enable operators to handle workload fluctuations, to keep the desired performance, with great agility and reduced costs. This procedure introduces a non-negligible delay in allocating or releasing virtual resources. Therefore, waiting until the system is overloaded or underutilized so as to scale resources up or down could negatively impact the users' Quality of Experience, or lead to inefficient resource utilization. In this vein, this paper proposes a novel and agile Dynamic Auto Scaling Algorithm for the Control Plane (CP) of the Long Term Evolution' (LTE) virtualized Evolved Packet Core (vEPC). The resources dimensioning stage of the algorithm is based on an original queuing model for the CP. To model the CP, we use an open network of G/G/m queues. We also provide expressions to derive the steady state transition probabilities of the queuing network. Finally, we validate the proper operation of our solution using accurate simulation tools.

Many ongoing research activities relevant to 5Gmobile systems concern the virtualization of the Evolved Packet Core (EPC) elements aiming for system scalability, elasticity, flexibility, and cost-efficiency. Virtual Evolved Packet Core (vEPC) will principally rely on some key technologies, such as Network Function Virtualization (NFV), Software Defined Networking (SDN) and Cloud Computing, for enabling the concept of Mobile Carrier Cloud. The key idea beneath this concept, known also as EPC as a Service (EPCaaS), consists in deploying virtual instances (i.e., Virtual Machines or Containers) of key core network functions (i.e., Virtual Network Functions - VNF), such as the Mobility Management Entity (MME), Serving GateWay (SGW), and Packet Data network gateWay (PGW) over a federated cloud. In this vein, an efficient VNF placement algorithm is highly needed to sustain the Quality of Service (QoS) while reducing the deployment cost. Our contribution, in this paper, is to devise an algorithm that derives the optimal number and locations of vEPC's virtual instances over the federated cloud. The proposed algorithm is based on coalition formation game, wherein the aim is to build optimal coalitions of Cloud Networks (CNs) to host the virtual instances of the vEPC elements. The obtained results clearly indicate the advantages of the proposed algorithm in ensuring QoS given a fixed cost for vEPC deployment, while maximizing the profits of cloud operators.

The existing conventional mobile networks are not flexible: if there is a new service, it unfortunately cannot be integrated automatically. Their traffic routing is not optimal and users' traffic is forwarded to the core network without considering the optimal path. This causes high latency to access the desired service, and the use of resources is inefficient. This has motivated the evolution towards 5G. The 5G vision consists of managing highly dynamic network slices and provisions networks in an as-a-service fashion. In this vein, to answer to the elasticity and low-latency specifications of the upcoming 5G services, the optimal placement of Virtual Network Functions (VNFs) must overcome the non-uniform service demand and the irregular nature of the underlying network topologies. This paper addresses this issue by mapping the non-uniform signaling messages to a new uniform environment, namely, the canonical domain, whereby the placement of core functions is more feasible and efficient. This is carried out by using Schwartz-Christoffel conformal mappings. The conducted experimentation shows the efficiency of our approach, compared to some baseline approaches, in the virtual resource allocation (i.e. Virtual CPU, Virtual DISK) and that is in terms of reducing the overall cost, end-to-end delay and number of activated Virtual Machines (VMs; virtual resources in general).

Many ongoing research activities relevant to 5G mobile systems concern the virtualization of the mobile core network, including the evolved packet core (EPC) elements, aiming for system scalability, elasticity, flexibility, and cost-efficiency. Virtual EPC (vEPC)/5G core will principally rely on some key technologies, such as network function virtualization, software defined networking, and cloud computing, enabling the concept of mobile carrier cloud. The key idea beneath this concept, also known as core network as a service, consists in deploying virtual instances (i.e., virtual machines or containers) of key core network functions [i.e., virtual network functions (VNF) of 4G or 5G], such as the mobility management entity (MME), Serving GateWay (SGW), Packet Data network gateWay (PGW), access and mobility management function (AMF), session management function (SMF), authentication server function (AUSF), and user plane functions, over a federated cloud. In this vein, an efficient VNF placement algorithm is highly needed to sustain the quality of service (QoS) while reducing the deployment cost. Our contribution in this paper is twofold. First, we devise an algorithm that derives the optimal number of virtual instances of 4G (MME, SGW, and PGW) or 5G (AMF, SMF, and AUSF) core network elements to meet the requirements of a specific mobile traffic. Second, we propose an algorithm for the placement of these virtual instances over a federated cloud. While the first algorithm is based on mixed integer linear programming, the second is based on coalition formation game, wherein the aim is to build coalitions of cloud networks to host the virtual instances of the vEPC/5G core elements. The obtained results clearly indicate the advantages of the proposed algorithms in ensuring QoS given a fixed cost for vEPC/5G core deployment, while maximizing the profits of cloud operators.

The evolution towards 5G consists of managing highly dynamic networks and making decisions related to the provisioning of networks in an as-a-service and cost-aware fashion. This is translated by 5G verticals that are dedicated to specific services, applications, or use cases fulfilling the constant demand of vertical industries. In this vein, to achieve the high-level goals defined by operators and service providers, and to answer to the elasticity and low-latency specifications of the upcoming 5G mobile system, the optimal placement of virtual network functions must cope with the non-uniform service demands and the irregular nature of network topologies. This paper addresses this issue by mapping the non-uniform distribution of signaling messages in the physical domain to a new uniform environment (i.e., canonical domain) whereby the placement of core functions is more feasible and efficient by means of Schwartz-Christoffel conformal mappings. The experimentation results, compared to some baseline approaches, have proven the efficiency of the conformal mapping based placement in allocating the virtual resources (i.e., virtual CPU and virtual storage) with regard to the optimal end-to-end delay, cost and activated virtual machines. Another interesting contribution is that all placement decisions are based on a realistic spatio-temporal user-centric model, which defines both the mobility of user equipments and the underlying service usage.

To support the much desired ultra-short latency of 5G mobile systems, many micro-data centers will be deployed in the vicinity of mobile users, defining a distributed edge cloud. Over this edge cloud, it is important to create optimal network slices to support different 5G verticals. Optimality is defined in terms of cost efficiency and QoS support. Therefore, it is important to understand the behavior of mobile users in terms of mobile service consumption. In this paper, we present, on one hand, a tool for developing a spatio-temporal model of mobile service usage over a particular geographical area. This tool will help to define the behavior of mobile users in terms of mobility patterns and mobile service consumption. On the other hand, based on this tool, we present a benchmark of some interesting Virtualized Network Functions (VNF) placement algorithms, among them our enhanced version of the predictive placement strategy. The comparison is based on data overload, overload of Virtual Machines (VMs) and QoS.