Linear Programming Lower Bounds for Minimum Converter Wavelength Assignment in Optical Networks

In this paper, we study the conflict-free assignment of wavelengths to lightpaths in an optical network with the opportunity to place wavelength converters. To benchmark heuristics for the problem, we develop integer programming formulations and study their properties. Moreover, we study the computational performance of the column generation algorithm for solving the linear relaxation of the most promising formulation. In many cases, a non-zero lower bound on the number of required converters is generated this way. For several instances, we in fact prove optimality since the lower bound equals the best known solution value.

Citation

ZIB Report 04-41, http://www.zib.de/ An extended abstract will be published in the proceedings of the International Network Optimization Conference, INOC 2005, Lissabon, Portugal.

Article

Download

View Linear Programming Lower Bounds for Minimum Converter Wavelength Assignment in Optical Networks