Traffic grooming in unidirectional WDM ring networks using genetic algorithms

Abstract

We propose a genetic algorithm (GA) with ( μ+ λ)-strategy to solve the traffic grooming problems in unidirectional SONET/WDM rings with arbitrary asymmetric traffic requirements. The objective is to optimally assign calls to wavelengths in order that the total number of SONET add/drop multiplexers (ADMs) is minimized while as few wavelengths as possible are used. We derive the theoretical lower and upper bounds on the numbers of both ADMs and wavelengths and summarize the general properties of the asymmetric traffic requirements in the ring. In applying GAs to solve the problem, we develop an order-based representation for the chromosome, introduce the concept of adaptive chromosome and propose a first fit approach incorporated with a greedy improvement algorithm to decode it. The order-mapped crossover and the inversion mutation are used to produce offspring. We also introduce the concept of the weighted average traversing length of each traffic and investigate the dependence of the average number of nodes at which per wavelength drops on both the weighted average traversing length and the average traffic requirement between each pair of nodes. Computer simulation results show that the algorithm proposed in this paper can significantly reduce the number of ADMs in solving the arbitrary asymmetric traffic grooming problems. A digital example shows that all the calls in the ring are nearly neatly paired after grooming.