A Matheuristic for a Telecommunication Network Design Problem with Traffic Grooming

This paper addresses a network design and traffic grooming problem arising in optical telecommunication networks that are based on wavelength division multiplexing. Given a set of nodes and a set of traffic demands between these nodes, the network design and traffic grooming problem (NDGP) consists of installing a minimum number of lightpaths between the nodes and of routing the demand on the lightpaths while respecting capacity constraints. We introduce a mathematical formulation of the NDGP as well as a hybrid algorithm capable of finding high quality solutions in short computing times. The proposed algorithm uses linear and mixed integer programming as slave methods and embeds them into a tabu search heuristic. Computational results and comparisons with an existing heuristic from the literature show the effectiveness of the proposed algorithm. Further analyses also show the efficiency of the neighborhood structure and of its evaluation technique.