Wireless Sensor Networks, Distributed Algorithms Based On Clustering Analysis

Wireless Sensor Networks (WSN) is becoming one of the most promising technologies in recent years. Many WSN applications require that sensor streams should be annotated with the physical locations of the corresponding nodes. Positioning without the aid of global positioning system (GPS) is crucial when GPS becomes impractical due to the limitations of cost, power, form factor etc.This thesis proposes a new distributed localization algorithm, Clustering-based Anchor-free Localization, brevity, CAFL. The basic idea of CAFL is to divide the whole sensor network into a number of clusters, in which the nodes calculate and refine their relative coordinates in parallel so as to set up the local coordinate frames. Furthermore, all local coordinate frames are reoriented so that the whole network converges to a single global coordinate system. The proposed algorithm employs the clustering technology to remove the limitation of the AFL algorithm in scalability of the WSN's size.Finally, comparative simulation studies are conducted in terms of a variety of node densities, range errors, network sizes. The experimental results show that CAFL is a robust localization algorithm and is significantly scalable, especially for the large size networks.