Research On Neighbor Information Based Localization Algorithm

Localization is a key issue in wireless sensor networks (WSN). Many applications ofWSN are based on location information. The location accuracy impacts on the servicequality and the performance of the entire WSN directly. However, the vast majority oflocalization algorithms for WSN are based on theoretical propagation model, and theperformance can not be verified in a real environment. Therefore, this paper aims to studythe characteristics of the real propagation model. Then design a localization algorithmwhich is able to be implemented in the real environment.The paper first designs a one-way propagation experiment in the case of simple andcomplex environments by the Mica2experimental platform. The characteristics of thewireless propagation model in real environment are obtained through the analysis ofexperimental data. We get the statistical property that path loss factor is not constant, butsubjects to the Weibull distribution. And by use of experimental data, we model thewireless propagation model of real environment. Secondly, we study a power-weightedcentroid localization algorithm after the propagation characteristics have been studied. Theweight function quantifies the impact of the location of anchor node to the estimatedlocation of the unknown node. The weighted exponent of the weight function isdetermined by the simulated annealing method. Then, we proposed a kind of neighborinformation based distance measurement, and the distance measurement has a good linearcorrelation with the real physical distance by use of the receive signal strength andneighbor sequence. Finally, we join the distance measurement with the power-weightedcentroid localization algorithm. That is the power weighted centroid localization algorithmbased on neighbor information. The algorithm is verified through simulation and practicalexperiment. The results prove that the algorithm can reach better localization accuracy inreal environment.