Study on efficient design and scalable control for wavelength-routed transport networks

Abstract

Unprecedented increases of internet traffic intrigue the fundamental research questions regarding network performances and resources as well as a sustainability of the Information and Communications Technology (ICT). State-of-the-art transport and IP networks are designed, operated, and managed by separately and this inefficient operation triggers researches about convergence networks. In this circumstance, in order to cope with the traffic explosion, we study on efficient design and scalable control for wavelength-routed transport networks. In the first part, we study on design and control problems for delay tolerant optical transport networks where store-and-forward functions improve a capacity of networks at the cost of queueing delay. We propose novel networks planning algorithms such as the bottleneckness centrality based buffer node location algorithm and analyze network performances as well as design rules by the exhaustive simulation studies. The second part investigates energy efficient IP-over-WDM dimensioning rules with traffic grooming algorithms. We study on philosophy and design rules of cross-layer networks by developing energy optimal formulations of traffic grooming. We further propose heuristic algorithms to achieve highly scalable networks with acceptable performance. Especially, we explore trade-off of energy consumption and propagation delay in the traffic grooming function. A centralized control plane in a software-defined networking paradigm intrigues classical but important research questions regarding performance and scalability of the networks. We propose heuristic RWA algorithms and investigate trade-off relation between a complexity and throughput in the algorithm by simulation studies. Proof-of-concept experiments of the proposed algorithm on the OpenFlow-based software-defined optical transport networking testbed demonstrate the validity and feasibility of the proposed researches in this dissertation.