Research Information System
COMPUTER AND INFORMATION SCIENCES - ISCIS 2004, PROCEEDINGS, vol.3280, pp.819-826, 2004 (SCI-Expanded)
The routing and wavelength assignment problem which is known to be NP-hard, in all-optical transport networks is considered. The present literature on this topic contains a lot of heuristics. These heuristics, however, have limited applicability because they have a number of fundamental problems including high time complexity, and lack of scalability with respect to optimal solutions. We propose a parallel hybrid genetic algorithm/heuristic based algorithm. Parallel genetic algorithms represent a new kind of meta-heuristics of higher efficiency and efficacy thanks to their structured population and parallel execution. The hybrid algorithm presented uses an object-oriented representation of networks, and incorporates four operators: semi-adaptive path mutation, single-point crossover, reroute, and shift-out. Experimental results of the test networks make clear that, when the network cost depends on heavily wavelength assignment, the proposed parallel GA/Heuristic hybrid approach provides promising results.