Espace membre

Cet espace est dédié aux étudiants, aux enseignants et au personnel administratif de l'école


Mot de passe oublié?



Doctorate thesis defense of Inès Raïssa Djouela Kamgang

Doctorate thesis defense on March 13th 2021 at 10H00 ,in Sup’Com Amphitheater Ibn Khaldoun.

Entitled :Optimization Models for Slice-Based and QoS-Enabling Service Orchestration Mechanisms in NFV

Presented by :Inès Raïssa Djouela Kamgang


President :

Sofiane Cherif

Professor at SUP'COM, Tunisia




Reviewers :

Arref Meddeb

Professor at ENISo, Tunisia


Kaouthar Sethom

Professor at ENICarthage, Tunisia


Examiner :

Slim Rekhis

Professor at SUP'COM, Tunisia




Supervisor :

Nabil Tabbane

Professor at SUP'COM, Tunisia


Offering a diversified, QoS-aware, and up to date panel of low-cost services remains the best means of customer loyalty for Telecommunications Service Providers (TSP). However, the proliferation of types of networks (2G, 3G, 4G, and already 5G) has so far had serious repercussions on the budget of operators since each new generation of network brings new proprietary physical equipment, and therefore new challenges in terms of staff training in the use of this equipment, maintenance etc. This fact, which generates high capital and operational charges (CAPEX and OPEX) and therefore high losses for the operator, makes it difficult to comply with the above-mentioned requirements.

Network Function Virtualization (NFV) is an emerging concept, which provides the required flexibility of the network to fulfill the mentioned challenges by reducing TSP’s capital and operational costs. The idea is to decouple the network functions from dedicated hardware equipment’s on which they lay. Thereby, a service which initially consisted of a set of physical functions linked together by physical links is transformed into a set of logical network functions named Virtual Network Functions (VNFs), linked together by logical links.

To formalize the NFV concept, the European Telecommunication Standard Institute (ETSI) has standardized a Framework which is the basis for the materialization. However, to achieve the expected gains, it is imperative that the problems hitherto not addressed in the development of that framework are addressed. These include issues related to the reduction of energy consumption, security, management, orchestration, and resource allocation. In this thesis, we have focused on joint orchestration and dynamic resource allocation issues.

This thesis aims to propose a Slice-Based, QoS-Enabling and Decentralized new NFV framework inspired by the one normalized by ETSI, that fulfills the user's requirements and enables the “pay as you go” concept. It also aims to propose efficient algorithms which could achieve an efficient end-to-end service delivery by meant of dynamic end-to-end resource allocation. To achieve this, we first present an overview of resource allocation and orchestration in NFV. We afterward present the new framework and its main building blocks, and we detail the service execution process in the framework. We then propose a heuristic to tackle the service chain composition issue, which is a subphase of the service chaining process, first stage of NFV resource allocation. Finally, we propose two heuristics to address the second and the third stages of NFV resource allocation, namely service function mapping and dynamic scheduling. Our dynamic scheduling approach performs better than other scheduling approaches (MINMIN, HEFT, Round Robin) while respecting the requirements of NFV-based networks.

Keywords :

Network Function Virtualization (NFV), Dynamic Resource Allocation, Service Function Chaining, Service Function Mapping, Service Function Scheduling, Orchestration, Optimization, QoS, “pay as you go”.