A heuristic method for clustering a large-scale sensor network

We present a new heuristic method for a clustering problem of sensor networks. The heuristic method is using the unca- pacitated facility location problem formulation for the clus- tering problem of sensor networks. It is an iterative method based on the Voronoi diagram. We also propose a parallel version of the heuristics to reduce the time to obtain a solu- tion. The proposed algorithms are investigated for the qual- ity of their approximate solutions and computational time to obtain them. By comparing the approximate solutions to the exact solutions for examples of one hundred sensors, we found that the quality of the approximate solutions is almost the same as that of the exact ones. The computational time to obtain the approximate solutions is a thousandth of that of obtaining the exact solution. For examples of ten thou- sand sensors, the computational time to obtain a solution is about 9.1 seconds by the sequential algorithm and about 6.0 seconds by our parallel algorithm with six computers.