Localization Using Self-Organizing Maps Based on Spring Relaxation

This paper proposes a self-organizing positioning algorithm of wireless sensor network based on constraints. The whole sensor network is considered as a spring network which is constrained by the distances between nodes, and combining with the topology preservation of self-organizing maps to achieve localization. The novelty of the paper is to introduce the concept of spring into self-organizing maps. An elastic spring is assumed between the nodes which are connectivity, and the nodes perform tensile and compression adjustment to equilibrium state according to the difference of the current and actual distances. The entire adjustment process is divided into coarse tuning and fine tuning which uses the constraints of the neigh boring nodes to improve the accuracy. Experiment shows that the proposed algorithm is with high positioning accuracy, good stability.