Designing B-ISDN network topologies using the genetic algorithm

The topology design of a B-ISDN network is addressed. We model the topological planning as a nonlinear mixed-integer programming problem. The genetic algorithm, an effective optimization method, is applied to this problem. Since the randomness of the generic algorithm cannot guarantee the biconnectivity requirement in the topologies generated by the genetic algorithm, we propose an algorithm to make all topologies at least biconnected while increasing the overall cost of the topologies the least. The result for a 20-node test case is presented and it is shown that the algorithm we propose has a very good convergence property.