Research On MDS-based Localization Algorithm Of Wireless Sensor Network

Wireless Sensor Network is a intelligent self-organizing network system whichIntegrated with computer technology, transduction technology and network technology.Recently, WSN has been widely used in human life. It can perceive and deal with variouskinds of data and information which exchange information between man and the physicalworld. Localization provides location services for the application layer, the transmissioncontrol layer and other layers. Location information of the nodes have a very importantrole in the application of wireless sensor networks, it has become a hot research topic ofWireless Sensor Networks. So, the research and design of high-precision, low-power,robust positioning technology has great significance.Multidimensional Scale (MDS) is a technique to analyses data, deriving frompsychology and psychophysics. It can maintain similarity when descends dimension ofdata. The localization algorithm which based MDS, only need little anchor nodes can stillachieve a high positioning accuracy, and the algorithm can be applied to based ranging orfree-ranging environment. So, this paper is intended to study localization algorithm whichbased on MDS.First, this article studies the WSN’s structure, feature and applications, and elaboratessome related measurement technique, analyses the criterion of localization algorithm. Then,the problem of poor adaptive ability on typical ranging model was analyzed. Through thequantitative relationship between the edges and adjacent information of anchor nodes, theadaptive ability of ranging model has been improved. Then, a detailed analysis of theproblem of MDS-MAP(P) have been proposed. In order to solve this problem, adensity-based clustering localization algorithm called MDS-MAP(DB) is proposed. Thealgorithm based on node density clustering algorithm to divide network into severalclusters, its size is adjusted by the common node. The distance of non-adjacent nodes inclusters is estimated by the triangle inequality method. New clustering algorithm does notrequire each node running the clustering; only the cluster head node executes the algorithmof MDS and combination. This kind of cluster algorithm not only reduces thecomputational complexity but also guarantees the public the number of nodes. Comparedwith the shortest path method, the triangle inequality method reduces the error of rangingmeasure and improves the positioning accuracy of algorithm. For the relative coordinates,iterative optimization is used to get a minimum value of error between the estimated coordinates and the true coordinates of nodes and improves the positioning accuracy.Finally, MATLAB as a simulator is used to realize the MDS-MAP(DB). With differentanchor nodes and connectivity, Simulation demonstrates that the improved localizationalgorithm has higher precision than traditional MDS-MAP algorithm.