Parallel island-based estimation of distribution algorithms for wireless network planning

Wireless network planning is an important problem to the communication companies. To solve this problem, many models were proposed, which overlooked different profits on the same acreage in different regions. Genetic algorithms (GAs) and similar algorithms were put forward by many papers based on those models. However, those models cannot reflect the companies' aims that make maximum profit with minimum cost. In this paper, we put forward a novel model with profit weight and penalty factor and design a new simulation experiment about the model. In order to gain the optimal solution quickly, a kind of novel algorithms - parallel island-based estimation of distribution algorithms (PEDA) - is proposed based on Larranage and Lozano (2001). Through simulation tests, we find that our algorithms outperform other stochastic heuristic search algorithms.