Solving the survivable network design problem with search space smoothing

In this paper, we propose heuristics for solving the survivable network design problem(SNDP) which is known to be NP-hard. In the SNDP, we want to obtain a minimum cost network design that satisfies survivability constraints given by the number of node-disjoint paths for each pair of nodes of a given graph. For solving the SNDP, we propose 2 heuristics that use a local search method with search space smoothing. The search space smoothing generates a series of smoothed search spaces that approximate the original search space. The smoothed search space has fewer local minima than the original search space and if we solve the problem with the smoothed search space, we can easily approach to the global minimum of the original problem. The results of computational experiments show the efficiency of the proposed heuristics.