Research On Unified Service Provisioning Platform And Service Discovery In Next-Generation Networks

Next-Generation Network (NGN) is the IP-based converged network, which is capable of providing all kinds of services and applications. IP Multimedia Subsystem (IMS) has become the common basis to establish the service provisioning platform for NGN. It is necessary to study the combination of IMS and other service provisioning approaches for achieving the unified service provisioning platform in NGN. In parallel, the convergence at the service layer of NGN has not only enabled a variety of service types, but also greatly augmented the quantity of services. Due to users' movement among different networks, it is a very crucial research topic to find out the best one which meets the user's requirements from the enormous services.This thesis is to study the unified service provisioning platform in NGN through integrating IMS and Peer-to-Peer (P2P) technologies, as well as the theoretical modeling and analyses of service discovery architectures. The main contributions of this thesis are as follows.1. A unified service provisioning platform in NGN based on the seamless integration of IMS and P2P technologies. We illustrate the integration of IMS and P2P from tow aspects. First, the IMS-based P2P overlay is proposed. Making use of IMS functional entities, it is achieved to create and maintain the proposed IMS-based P2P overlay. By carefully comparing the proposal with the existing P2P overlays, the advantages of our IMS-based P2P overlay is exploited. Secondly, P2P is applied to speed up the information lookup within IMS. Simulation results show that the introduction of P2P into IMS can significantly improve the efficiency and scalability when performing the lookup. 2. An analytical model for centralized service discovery architecture. The service directory is mapped onto an M/G/c/c queuing system. The frequency to transmit the service announcements directly influences the availability of information hold by the directory, as well as the network traffic for service discovery. Thus, we further study the periodical service announcements from an individual service provider. And, the expressions for success rate, network traffic and cache utilization are derived. Based on the thorough performance evaluation with numerical simulations, an optimization approach is proposed. Selecting the optimal update interval for service providers not only ensures the success of service discovery, but also lowers the requirements on both the network and the directory to facilitate the adoption of centralized service discovery architecture.3. An analytical model for semi-centralized service discovery architecture. By elaborating the evolution from centralized service discovery architecture towards semi-centralized one, we determine the necessary extensions to the model for centralized service discovery architecture. The directory is still mapped onto a queuing system, and it is very important to study the service announcements from an individual service provider. Moreover, we consider the impacts of the directory's change on the model. Based on the numerical results, a novel semi-centralized service discovery protocol is proposed and the network simulations are conducted. Even if our proposal shows a little degradation in success rate and response time, the network traffic is saved by at least 30%. It is really a worthy trade-off in wireless environments with limited bandwidth resources.4. A generic model for two basic service discovery architectures. Based on the in-depth comparisons between two basic service discovery architectures, we determine the foundation to explore the generic model. Then, the generic model is illustrated and the generic expressions for various performance metrics are derived. We further exploit the differences of two basic service discovery architectures in each performance aspect. Since service discovery architectures use Time-To-Live (TTL)-based control mechanisms to save the bandwidth, we also model the impacts of the TTL's value. Making use of network simulations, the proposed model is validated. And, we obtain the appropriate approaches to separately set the TTL's value for two different service discovery architectures.