Déploiement d'IMS et provisionnement de la QoS dans le nuage informatique

Abstract

IP Multimedia Subsystem (IMS) is revolutionary technological framework that changed the way we communicate based on standardized architecture that provides a variety of multimedia services such as telephony, text, picture, video , Internet or a combination of them over one IP network. This architecture is adopted by most of the operators over the world as a part of their networks and enabled a reach market with standardized services that are scalable, available and dependable. However, with the rapid development of information technology and communication (ICT), telecom operators are facing many challenges in understanding the requirement of the mobile phone market and customers include more new services and bring new opening IT solutions concurrent with changes in the market. These challenges are due particularly to the complex architecture of the operators networks and the hard process of services deployment, the lack of value-added services and the huge CAPEX and OPEX required for the integration of these new services, the rapid and the continually growth of network services demand and devices connected to the operators networks which is linked to the increase of the number of subscribers, also the limited capacity of the existing networks equipments to handle all the users request with with the same telecom grade. Today, cloud computing is changing the ways of providing services over Internet Protocol. With the attractive characteristics of Cloud Computing, on-demand network access, shared pool of configurable computing resource comprising networks, severs, storages, applications and services that can be quickly provisioned, additionally the scalability and the elasticity that can be managed with a less user interaction. It’s bringing a new market opportunity to the operators with the migration of the telecom applications and services to the Cloud Computing environment. In this thesis we aim firstly to deploy the Telco application IMS to the Cloud ecosystem. Through IMS cloudification, operators will have more flexibilities and facilities to deploy new or existing Telco’s services and to bring new value added to their network with a better performance and lower cost and time. Given that the quality of service (QoS) is one of the key elements related to the IMS and it is very rational for real-time services provided by this system. The lack of computing resources degrade the QoS and affects the user perception and satisfaction of the service. From this controversy we proposed a QoS provisioning model for the deployed IMS architectures in the cloud computing environment. The proposed model "IMS QoS Provisioning" aims to automate the provisioning process of the IMS resources to provide a better QoS to the end users. In order to validate the IMS cloudification, we used the open source IMS product. Thus, we deployed two testbeds architectures in the Cloud using an open source IMS solutions : Open Source IMS Core (FOKUS, 2006) and Clearwater IMS (Metaswitch, 2013). As an infrastructure we used an open source Openstack (Rackspace, 2013) as our Infrastructure as a Service (IaaS) and our solutions are deployed in the private cloud of the Synchromedia using Ericsson Blade Servers. To compare the performance of the cloud IMS architectures with the fixed IMS we implemented a third testbed where we used Open Source IMS Core solution (FOKUS, 2006). The functional testing of the Cloud IMS testbeds has been conducted successfully as well for the traditional solution using several free SIP based software clients. The performance results show that both the virtualized IMS solutions give similar if not better performances than the fixed IMS solution deployed in physical hardware. However, Cloud Clearwater IMS has always given a better performance than the other two architectures. General measurement showed the variation of the QoS metrics RRD and SRD delay is linked to the number of the user connected to the operators’ network as well the virtual resources usage (CPU and memory) of IMS hosted virtual machine. By increasing the number of phone calls to the IMS network, we will have an increase in the latency of the control plan metrics (RRP, SRD, IMD) and data plane (RTP delay, Inter arrival jitter). This increase in the number of users procreate a semi-linear growth of the resources required by each virtual machine.

Date	10 Feb 2015
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Mohamed Cheriet (Supervisor)