Support des applications multimedia dans les reseaux de prochaine generation

The first two works deal with the issues of the Fixed-Mobile Convergence (FMC). This convergence is blurring the distinction between mobile and fixed networks. Among other things, it allows a user to have access to its services, both on the cellular network (Long Term Evolution (LTE), for example) as well as on a local network (Wireless Fidelity (WiFi), for example). This is usually accomplished by equipping the user with a device with that can connect to both networks. The first issue raised in this thesis is about the decision of when to execute a handover. The two most popular mobility protocols, Mobile IP (MIP) and Proxy Mobile IP (PMIP), approach this problem with diametrically opposed views. With the first protocol, the decision is made by the user and his device. With MIP, many signaling messages are sent, wasting valuable network resources, especially at the radio access, which is the most precious part of the network. Furthermore, by not involving the network, the decision taken by the device will not be necessarily optimal. For these reasons, the PMIP protocol was proposed. Its approach is the opposite of that of MIP, the mobility being managed entirely by the network. By doing so, the device is actually shielded from any aspect of the mobility, and is fooled into thinking that it's always in its home network. This is possible by introducing new nodes in the networks that act on its behalf, which eliminates all signaling on the radio link. In addition, since the network is usually better suited to make the right decision, because of the additional information it holds, the mobility is optimized. However, the big issue that arises is that, without the intervention of the terminal, it is impossible to detect all the handover possibilities. Thus, PMIP is not a protocol that is well suited to deploy FMC.;The first article, entitled "Client-Based Network-Assisted Mobile IPv6", therefore addresses this problem by proposing a new protocol based on Mobile IP v6 (MIPv6), in which we introduce the involvement of the network. The result is a hybrid protocol that draws upon the strength of MIPv6 and Proxy Mobile IP v6 (PMIPv6). To accomplish this, two steps were required. The first consisted of a complete overhaul of the MIPv6 protocol, as in its current state, it was near impossible to make any modifications. The result is a much more lightweight protocol which provides only basic functionality. Other features were separated into modules. In the second step, we proposed a new module, which introduces a new node in the network that can handle the terminal mobility. Thus, the collaboration of the terminal and the new node reduces the signaling messages and optimizes the decisions for handing over, while still offering support for FMC.;The second issue that this thesis tackles is the seamlessness of a handover between two networks. A handover is deemed seamless if it does not cause any disruption to the user's services. The applications targeted by our work are multimedia applications operating in real-time, such as IP TeleVision (IPTV) and Mobile IP TeleVision (MobileTV) (in live television mode). The problem with these applications is that when a vertical handover occurs (in the case of FMC, for example), the connection is lost and must be re-established.;The second article, entitled "Seamless handover for multicast Mobile IPv6 traffic", proposes a solution to this problem. This is accomplished by introducing a new node in the network, whose role is to buffer the lost packets while the handover is occurring. Thus, when the device reconnects, it is able to recover these packets. The user application is therefore able to proceed without interruption.;The third issue addressed in this thesis focuses on the planning of access networks, to support the high bandwidth required by multimedia applications. For the FMC to be successful, it is necessary that the local network supports the bandwidth requirements. New hybrid optical technologies have been proposed.;In a new environment, there is no infrastructure to reuse. Therefore, the consensus in such a scenario is that the best technology to deploy is the one that only uses fiber optic links, as it offers the best rates and the greatest scalability.;The planning must be dynamic and take into account the changing nature of customer demands. The third article, entitled "Dynamic Greenfield Fiber to the Home Planning", proposes a dynamic model for the network planning problem of fiber optic networks. The result is a linear integer mathematical model, which takes input parameters, such as customer demands, and produces a planning that minimizes the total cost of the network, over all of the phases. (Abstract shortened by UMI.).