Abstract

Multimedia Conferencing is the real-time exchange of media content (e.g. voice, video and text) between multiple participants. It is the basis of a wide range of conferencing applications such as massively multi-player online games and distance learning applications. For faster development as well as cost efficiency, developers of such conferencing applications can use conferencing services (e.g. dial-in audio conference) provided by third-parties. However, the third-party service providers face several challenges with respect to conferencing service provisioning (i.e. service development, deployment and management). One challenge is mastering complex low-level details of conferencing technologies, protocols and their interactions. Another challenge is resource elasticity. Number of conference participants varies during runtime. So resource utilization in an elastic manner is a critical factor to achieve cost efficiency.

Cloud Computing can help tackle these challenges. It is a paradigm for swiftly provisioning a shared pool of configurable resources (e.g. services, applications, network and storage) on demand. It has three main service models: Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS) and Software-as-a-Service (SaaS). Using a PaaS, service providers can provision conferencing services easily and offer them as SaaS. Nonetheless, cloud-based provisioning of conferencing services still remains a big challenge due to the shortcomings of existing PaaS.

In this thesis, a PaaS architecture for conferencing service provisioning is proposed. It is based on a business model from the state of the art. It relies on conferencing IaaSs that, instead of VMs, offer conferencing substrates (e.g. dial-in signaling, video mixer and audio mixer). The conferencing PaaS enables composition of new conferences from substrates on the fly. Moreover, it provides conferencing service providers, who are experienced in programming, with high-level interfaces to abstract the internal complexities of conferencing. In order for PaaS to scale ongoing conferences elastically, an algorithm is also presented in this thesis. The conferencing PaaS is prototyped and performance measurements are made. The proposed algorithm’s performance is also evaluated.