Research On Localization Algorithms For Wireless Sensor Networks

Wireless sensor networks, a novel technology about acquiring and processing information, have been applied to a wide variety of applications such as military, medical, environmental monitoring. It is a pioneering research field which attracts the most attention from around the world. However localization is a fundamental problem in the wireless sensor networks. For the variance of environment, a localization algorithm always has to fight against the interference from environment during its course of obtaining a precise localization result, which is the main goal of all localization algorithms.Multidimensional scaling (MDS) is a data analysis technique that transforms proximity information into a geometric embedding. MDS is well-suited to node localization in wireless networks, where the task is to use the correlative information of nodes to determine the coordinates of nodes in a 2D or 3D space.Based on the research of MDS-MAP localization algorithm (a centralized algorithm), a new MDS-based Distributed localization algorithm is presented. First each node collects the distance information to its neighbor nodes in a broadcast way.Each node generates a local distance matrix with the distance information. Then a local relative coordinates system is build using Multidimensional scaling technique. Second all the local coordinates system are patched together to form a global map. Finally the global maps are transformed to absolute coordinate system with the anchor nodes whose absolute coordinates and global map coordinates are both known. The distributed methods are suitable for most kinds of topology networks.During the course of produce local node distance matrix, a new Euclidean distance estimates method from Euclidean localization algorithm is introduced which makes the estimate distance between nodes pair more precise. Besides, a more effective coordinates transforming method is used to replace the original one. Through these two measures the localization error and algorithm complexity are reduced further.Simulation experiments have been carried out to verify and test the new algorithm. Simulation results show that MDS-based Distributed algorithm has a better localization accuracy which is little dependent on the number of anchor nodes. Also, it can be applied to all kinds of topology networks.