| In this thesis, a detailed and comprehensive study is presented on the Universal Mobile Telecommunications System (UMTS) network planning problem. This problem has been shown to be NP-hard. Therefore, approximate algorithms are necessary to build planning tools. Three planning tools, based respectively on genetic algorithm, simulated annealing and a novel cooperative method, are designed and implemented to solve the global planning problem of UMTS networks.;Using the optimal solutions as references, numerical results are compared amongst the proposed planning tools and a previously designed tool based on tabu search. The cooperative method shows its superiority over the three other planning tools with 90 percent confidence that the true mean solution gap from the optima is within the interval of [0.01%, 0.33%I. Moreover, this solution closeness to optima is not necessarily accompanied with long computation time. These observations make the cooperative method more appropriate for global planning of UMTS networks. |
