IP network planning for realtime services with statistical QoS guarantees

This dissertation investigates various areas of IP network planning for realtime services. These areas are traffic characterization, capacity assignment, robust network dimensioning, and network planning tool development. In regards to traffic characterization, various clients of realtime services are tested in different situations and their traffic is characterized using known models. Motivated by the fact that hard quality guarantees incur extremely high costs, this dissertation exploits the issue of statistical guarantees and introduces two novel approaches for capacity assignment for interactive voice and video services, respectively. Subsequently, it proposes a statistical way that accounts for traffic demand variability in the context of network dimensioning to assure a sufficiently robust network. Finally, a software planning tool with a generic layered architecture is developed for converged IP networks.