A Cooperative Multilevel Tabu Search Algorithm for the Covering Design Problem

This work describes an adaptation of multilevel search to the covering design problem. The search engine is a tabu search algorithm which explores several levels of overlapping search spaces of a t - (v, k, λ) covering design problem. Tabu search finds good approximations of covering designs in each search space. Blocks from those approximate solutions are transferred to other levels, redefining the corresponding search spaces. The dynamics of cooperation among levels tends to regroup good approximate solutions into small search spaces. Tabu search has been quite effective at finding re-combinations of blocks in small search spaces which provide successful search directions in larger search spaces.