An innovative technique for backbone network design

An algorithm is proposed for solving the minimum-cost backbone network design problem. The method includes time delay as the performance criterion and multiple-hop communication paths. The algorithm combines a branch-and bound method with the algorithm of L.R. Ford and D.R. Fulkerson (1962). The branch-and-bound method is used to find the network topology. A depth-first search using a cost heuristic that is a linear function of the maximum and current depth in the tree is used. When a feasible topology is found, the Ford-Fulkerson algorithm is applied to find the optimal paths. The resulting network is retained if it is the current optimum cost network. The algorithm is compared with a branch-and-bound algorithm and is found to produce globally optimal solutions. For large, sparsely connected networks computation times are reduced by a factor of over one thousand.< >