| The ability of core networks to manage data transmission of increasing volume and variation is critical for the success of data-intensive and network-centric applications as they grow in both scale and complexity. Traditionally, static optical networks were the dominant transport for medium and long distance communication. However, these networks can no longer meet the needs of tomorrow's applications for higher bandwidth at lower cost. New dynamic optical networks greatly improve the reconfigurability of optical terminal systems and support unprecedented flexibility for high-traffic resource sharing.;However, managing dynamic networks poses challenging problems related to scale and traffic volume. Traditional analytical techniques, which rely heavily on simplification of network topologies and route choices, are insufficient to understand the significant performance differences implied by the subtle path preferences of dynamic routing algorithms.;This dissertation presents an integrated approach to efficient and robust resource management algorithms for on-demand data traffic on dynamic photonic switching networks. First, a resource management framework is proposed to consider both resource dimensioning and routing for optical opaque networks. I develop dimensioning and routing algorithms that are efficient to implement and robust to evolutions of traffic load, network topology, and scale. Second, using Poisson traffic assumptions, the thesis develops an opportunity cost model for analyzing threshold-based dynamic routing algorithms. The model is scalable and provides better congestion management than previous work. Third, applying the dimensioning technique developed in this thesis, I have several important findings on managing dynamically routed optical translucent networks, optical restorable networks, and multiple optical network domains. Finally, the thesis solves a combinatorial optimization provisioning problem for a dynamic wavelength service traffic model on an optical translucent network. This work is the first to evaluate the robustness of optical route rearrangement. New solutions are proposed to design an optimal dynamic service network with rerouting capability. |
