Resource Allocation in Multigranular Optical Network

Thesis Statement: Cost-effective switching and spectrum utilization efficiency have become critical design considerations in optical networks. This dissertation provides in-depth exploration of these important aspects, and proposes effective techniques for low-cost switching architectures and resource allocation algorithms to facilitate the adoption of optical networks in the near future.;The dramatic growth of Internet traffic brings challenges for optical network designers. The increasing traffic and bandwidth requirements mean that various resource allocation schemes to achieve different network design goals assume great importance. The general problem of resource allocation to lightpath requests is a challenging problem.;An emerging technology of flexible and more fine-grained grid through the use of Optical Orthogonal Frequency Division Multiplexing (OOFDM) allows fiber bandwidth to be more suitably matched up with application requirements, thereby making the network more elastic than the conventional Wavelength Division Multiplexing (WDM) optical networks. Despite the advances of employing OOFDM technology in elastic optical networks (EONs), imminent fiber capacity exhaustion due to the ever-increasing demands means that multiple fibers per link will be inevitable. While increasing the number of fibers boosts the capacity of networks, there is a price to pay for it in the form of increased number of switch ports / complexity of switches. The huge amount of traffic demands and thus high hardware requirements motivate multigranularity (such as wavebanding) to save costs in optical networks. This dissertation aims to tackle several types of resource allocation challenges in multi-granular optical networks to either improve the spectrum utilization or provide cost-effective switching techniques.;The first part of this dissertation presents our work on an important joint scheduling (or {em co-scheduling} of computational and network resources) problem in today's applications such as cloud computing. Such applications involve the processing of complex jobs consisting of several inter-dependent tasks executing on heterogeneous clusters of computing resources, which are interconnected by high-speed optical networks. Our co-scheduling algorithms take the constraints introduced by elastic optical networks into account and aim at either minimizing the makespan of a set of jobs in the static case, or minimizing the job blocking when jobs arrive dynamically.;The second part of the dissertation addresses problems on waveband switching in WDM networks. Grouping together a set of consecutive wavelengths in a WDM network and switching them together as a single waveband could achieve savings in switching costs of an optical cross-connect (OXC). A specific problem that we consider is the optimal Band Minimization Problem in WDM mesh networks to minimize the required number of switching elements while accommodating a set of traffic demands. Then we evaluate OXC node architectures in WDM networks with multiple parallel fibers. A hierarchical architecture that utilizes wavebanding and has lower complexity is compared with the conventional architecture in terms of the cost of the OXC node and the power consumption. Analytical models for computing the blocking probability of connection requests are proposed and validated.;In an effort to reduce the complexity of optical OXCs, a flexible wavebanding OXC architecture has been proposed recently. Elastic networking and flexible wavebanding introduce a new problem, namely, the routing, fiber, waveband, and spectrum assignment (RFBSA) problem. In the third part of the dissertation, a framework for solving the RFBSA problem in networks with multi-fiber links that can accommodate non-contiguous and non-uniform wavebanding is proposed. Then a joint banding-node placement and RFBSA problem to meet the network budget while maintaining good network performance is addressed.;The final part of the dissertation addresses dynamic Routing and Spectrum Assignment problem in multi-fiber EONs to further improve spectrum efficiency. We propose and evaluate novel schemes to minimize the demand blocking ratio of requests.