A Genetic Algorithm Approach to the Grooming of Dynamic Traffic in Tree and Star Networks with Bifurcation

Abstract

Traffic grooming is widely employed to reduce the number of ADM's and wavelengths. We consider the problem of grooming of dynamic traffic in WDM tree and star networks in this paper. To achieve better results, we used the bifurcation techniques to the grooming of arbitrary dynamic traffic in a strictly non-blocking manner in networks. Three splitting methods, including Traffic-Cutting, Traffic-Dividing and Synthesized-Splitting were proposed. A genetic algorithm (GA) approach based on these methods was proposed to tackle such grooming problems in tree and star networks. The performance of these algorithms was tested under different conditions in star and tree networks. Computer simulation results showed that our algorithm is efficient in reducing both the numbers of ADM's and wavelengths.