Title: Local search study of honeycomb clustering problem for cellular planning

Authors: Jean-Charles Creput, Abderrafiaa Koukam

Addresses: Systems and Transportation Laboratory, University of Technology of Belfort-Montbeliard, 90010 Belfort, France. ' Systems and Transportation Laboratory, University of Technology of Belfort-Montbeliard, 90010 Belfort, France

Abstract: We study a local search approach for a coverage problem in the plane, called Balanced Honeycomb Clustering Problem (BHCP), where a honeycomb mesh is used to build irregular hexagonal clusters that have to cover a fixed amount of points of a given data distribution. This problem has application to dimension cellular networks adapted to radio-mobile traffic. The local search approach uses dynamic application of fitness landscape penalties in order to exit local minima and improve performances, while eliminating overloaded cells. Results are given in comparison to best solutions known generated by an evolutionary algorithm. For a considerable reduction of computational time, about 3000% lower, we show that local search outperforms a population-based evolutionary approach.

Keywords: cellular networks; network dimensioning; adaptive mesh generation; evolutionary algorithms; mobile networks; local search; honeycomb clustering; cellular planning.

DOI: 10.1504/IJMNDI.2006.010816

International Journal of Mobile Network Design and Innovation, 2006 Vol.1 No.2, pp.153 - 160

Published online: 05 Sep 2006 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article