A K-means based firefly algorithm for localization in sensor networks

Abstract

Location information is essential and indispensable for various applications of sensor networks. The multilateration algorithm (MA) is a typical algorithm, which has been widely applied for localization due to its simple model and low requirement on hardware devices. Unfortunately, current methods to find the optimal solution in the MA show poor performance in localization accuracy and convergence. To solve this problem, a K-means based firefly algorithm (KFA) for localization in sensor networks is proposed. A weighted K-means method is designed to optimize the fitness function of the localization model. Then, an inertia weight factor is produced to improve the searching speed of solutions by reasonably adjusting the attractiveness of fireflies. In addition, a new updated solutions strategy is provided to improve the global searching capacity and avoid the local convergence. The simulation experiments demonstrate that the KFA presents better performance in convergence, stability and localization accuracy compared with other tested algorithms.