A New Approach for Network Topology Optimization

We consider the problems of designing netwrok topologies with minimum cost. In the first problem, basic optimum topology which guarnatees only one path from one to the other node, is considered. This problem is extended in the second part to include one redundant path between two nodes. These two problems are solved in this work in three stages. In the first stage, modified version of random feasible topology generation algorithm is used to generate initial feasible solutions. In the second and third stages, Link Reduction Algorithm and Branch-X-Change algorithms are used sequentially to get a locally optimal solution. These three stages are required to repeat with different orderings of the nodes to get a global optimal solution. We tested the above approach with two problems.