Research Of Self-Localization Algorithms In Wireless Sensor Networks

Since many applications of wireless sensor network depend on the location information of sensor nodes, how to obtain the accurate location has become the focus of attentions. The thesis focuses on self-localization algorithms.The weight centroid algorithm, which is range-free, has the advantages of low complexity and appropriate precision. But it performs well only with many anchor nodes. In this case, two new algorithms are introduced: the iterative weight centroid algorithm and the multi-hops weight centroid algorithm. The former uses the located nodes as anchors and controls the cumulate of error by some methods; the latter exploits the anchor nodes n hops away in the localization. Simulation results indicate that they perform well in few anchor nodes case.The AHLos algorithm, which is range-based, possesses high precision. Since the performance of AHLos algorithm is affected by range error, several improvements are proposed in this paper as follow: firstly, as the accurate distance between two neighbor anchors is known, we can use it to estimate the range error, thereby correct the measured distance between unknown nodes; secondly, different weights are given to anchor nodes, first-located nodes and latter-located nodes as their confident coefficients; finally, two conditions for located nodes upgrading to be anchor ones are presented in this paper. Simulation shows that the performance of improved algorithm is 10 to 20 percents better than the original one in high range error case. For the sake of reducing range error, this paper also focuses on a linear programming algorithm with triangle inequality constraints. We improve this algorithm by adding some triangle constraints so as to get a tighter solution. Each triangle referred above contains two anchor nodes. Simulation shows that the linear programming algorithm performs well in reducing range error and the improved one does better.