An ACO algorithm to design UMTS access network using divided and conquer technique

This paper studies the problem of planning UMTS (Universal Mobile Telecommunication System) access network. The aim is to determine the optimal number and location of radio network controllers (RNCs) and to find the connections of minimal cost between RNCs and radio base stations (RBSs) satisfying all the topological constraints. As the problem is NP-hard we propose a hybrid ant colony optimization (ACO) algorithm to tackle it heuristically. The main characteristic of the ACO algorithm is to perturb a saving-based greedy heuristic in its solution construction. We then use decomposition ants (D-ants) to enhance the efficiency of the algorithm. This is achieved by decomposing the master problem and solving only the much smaller sub-problems resulting from decomposition. Comparing with the previous results we will demonstrate through a number of test cases that our algorithms improve best previous results.