Optimization of fixed network design in cellular systems using local search algorithms

Search techniques such as genetic algorithms, simulated annealing, tabu search and random walk algorithms have been used extensively for global optimization. This paper presents an experimental analysis of the performance of these algorithms for the problem of designing the fixed portion of a cellular network. We first investigate the effect of various algorithm specific parameters on their performance. The algorithms are then compared to each other using criteria that ensure fairness. Under the given problem formulation and assumptions regarding the location of nodes in the network, we find that tabu search and genetic algorithms provide good, robust solutions.