The Multi-Dimensional Scaling Algorithm with Centroid as Reference Point for Location in Wireless Senso Network

Abstract

The multi-dimensional scaling(MDS) method is a kind of method used to analyze distance matrix or dissimilarity matrix to form data points in low-dimensional space. MDS algorithm uses the distance information between nodes in wireless sensor networks to estimate the relative location of nodes. Classical multi-dimensional scaling algorithm chooses any node in the network as reference point to estimate node location. With the increase of distance measurement error, the node position error estimated by this method also increases significantly. In order to improve the location performance, this paper proposes an MDS algorithm based on the centroid of all nodes in the network as a reference point, and derives the calculation formula of MDS algorithm based on centroid as reference point. Compared with the classical MDS algorithm, the simulation results show that the MDS algorithm base on centroid as the reference point has better positioning results.