Research Article
Service Provisioning under a Scheduled Traffic Model Using Light-trails in WDM Optical Networks
@INPROCEEDINGS{10.1109/BROADNETS.2007.4550459, author={Xubin Luo and Bin Wang}, title={Service Provisioning under a Scheduled Traffic Model Using Light-trails in WDM Optical Networks}, proceedings={4th International IEEE Conference on Broadband Communications, Networks, Systems}, publisher={IEEE}, proceedings_a={BROADNETS}, year={2010}, month={5}, keywords={}, doi={10.1109/BROADNETS.2007.4550459} }- Xubin Luo
Bin Wang
Year: 2010
Service Provisioning under a Scheduled Traffic Model Using Light-trails in WDM Optical Networks
BROADNETS
IEEE
DOI: 10.1109/BROADNETS.2007.4550459
Abstract
In this work, we study the problem of efficient service provisioning in light-trail WDM optical networks under a scheduled traffic model. In this model, a set of demands is given, and the setup time, teardown time, and the requested bandwidth of a demand are known in advance. When constructing lighttrails to accommodate scheduled demands, we consider two cases: static light-trail case and dynamic light-trail case. In the static light-trail case, a light-trail will not be torn down once it is set up. In the dynamic light-trail case, a light-trail can be torn down when it is not used by any scheduled demand, so that the resources for that light-trail can be re-allocated for other light-trails. We present Integer Linear Programming (ILP) formulations for this problem while considering the instantiation of static light-trails as well as dynamic light-trails to provision scheduled services. The optimization objective is to minimize the resources used in terms of the total wavelength-links used to accommodate a set of demands. Time efficient heuristic algorithms are then proposed for both cases. The solutions to the ILP formulations and the extensive simulation of the heuristic algorithms show that the use of dynamic light-trails significantly reduces the required resources to accommodate the same demand set. The resource reduction becomes more significant as the size of demand sets increases and the time correlation among demands becomes weaker. The heuristic algorithms for both cases are also shown to be very time efficient.
