| Emerging trends in scientific exploration are directing a steady evolution in unprecedented networking requirements. Optical wavelength-routed networks constitute the majority of the high-speed, low-latency technology configurations essential for supporting future communication. These systems enable parallel transmission of massive datasets on non-interfering wavelength channels. However, as scientific endeavor expands, the simple availability of multi-wavelength resources will fall short of sufficient. Rather, resources must be allocated intelligently, efficiently, and flexibly to bear the burdens of high-volume science.;Efficiently scheduling prospective optical traffic is among the most complex problems in the field of communications. Each demand must be allocated a collection of physical and spectral resources in order to provision a reservation - known as a lightpath - on the network, while simultaneously identifying an appropriate set of temporal resources to effectively schedule the solution. At scale, the construction of a comprehensive and efficient lightpath schedule is a juggling act of budgeting across a fairly broad set of resource-specific domains in aggregate.;This work introduces allocation opportunity among the synergistic resource domains via a novel family of lightpath scheduling enhancements called Lightpath Switching. Lightpath switching supports intermittent alteration to the lightpath solutions assigned to each reservation in an effort to expose potential for reducing resource fragmentation and increasing utilization efficiency. The purpose of this work is to systematically investigate the advantages and disadvantages of exposing resource provisioning to spectral-wavelength, spatial-routing, and terminal-destination Lightpath switching flexibility. These enhancements are applied in turn, as well as in combination, in order to furnish commentary on the relative efficacy, costs, benefits, and tradeoffs of supporting lightpath switching in various arrangement.;Each corresponding scheduling problem is be proven intractable, belonging to the complexity class NP-Complete. Optimal solution space modeling, heuristic approximation, and extensive quantitative and qualitative analysis is performed for a broad range of traffic inputs and network configurations in order to facilitate evaluation. Performance is examined at multiple cross-domain resource granularities, while opportunities to mitigate switching cost overhead are assessed. Findings indicate substantial efficiency gain, particularly via spectral lightpath switching. |
