Adaptive evolutionary algorithm for traffic grooming in wavelength-division multiplexing networks

Abstract

We study the traffic grooming problems in wavelength-division multiplexing (WDM) networks with arbitrary topologies (such as the ring or mesh WDM network), aiming to improve the throughput and minimize the blocking probability in the network. The grooming problem is formulated as an optimization problem in the network, and the optimizing solution consists of three parts of design: the virtual topology, the route of the lightpaths, and the assignment of the wavelengths to the lightpaths. An evolutionary algorithm, named adaptive immune evolutionary algorithm was proposed. The feature of the proposed algorithm is in introducing the adaptive parameters so as to avoid premature convergence and enhance searching efficiency toward its solution. Experimental results and comparison with two well-known algorithms show that the proposed algorithm has better resource utilization for random traffic grooming in WDM networks.