Solving large problem instances of the RWA problem using Genetic Algorithms

Abstract

This research work focuses on Integer Linear Programming (ILP) formulations for the Routing and Wavelength Assignment (RWA) problem in Wavelength Division Multiplexed (WDM) networks where end-users communicate with each other by establishing all-optical, non-interfering wavelength channels which are referred as lightpaths. The RWA problem is reducible to Graph Coloring problem in polynomial time and hence found to be NP-Complete. So, Soft Computing techniques along with Heuristic methods are applicable to solve large instances of the RWA problem. In this work, we propose new ILP formulations by imposing additional constraints to the objective function, thereby, establishing lightpaths which are immune to signal distortion and crosstalk. Furthermore, we develop constraints to make the lightpaths avoid creating loops while traversing between the Source-Destination node pair. After modeling the RWA problem as an Optimization problem, we focus on applying Genetic Algorithms (GA) to find a sub-optimal solution for the RWA problem.