Improving genetic algorithms for arbitrary traffic grooming in wavelength division multiplexing ring networks

Abstract

In wavelength division multiplexing (WDM) ring networks, subwavelength traffic streams can be elaborately arranged in wavelength channels to minimize the total cost due to the implemented wavelengths, electronic terminals, and so on. In this work, a genetic algorithm is proposed with a permutation-based chromosome representation and a selection method of the roulette wheel model to solve these traffic grooming problems, which can minimize the number of required electronic terminals, and keep high utilization of wavelengths. Some features of the algorithm are modified for high performance. To improve the algorithm further, different strategies are proposed to enhance the chromosomes in the decoding procedure. Computer simulations are performed with randomly generated traffic patterns to compare their effects and the results are analyzed.