B. Adenso-Díaz and M. Laguna
International Journal of Operations & Quantitative Management, vol. 7, pp. 105-118 (2001)
A new heuristic solution procedure for the solution of the rectilinear Steiner problem is proposed. The procedure is based on an innovative neighborhood structure that is explored using a pseudo-greedy approach. The probabilistic rules designed to induce search diversification are based on move attributes derived from the neighborhood structure. A Taguchi design of experiments is used to find the best settings for the search parameters associated with the proposed procedure. Extensive computational experiments show that the proposed procedure is superior to existing solution methods in terms of solution quality.
Full text
