Comparative study of RSS-based collaborative localization methods in sensor networks

Reliable localization is an essential building block of sensor networks. Many techniques have taken advantage of the received signal strength (RSS) measurement for location estimation in wireless sensor networks, since no special hardware implementation is required to measure RSS in almost all kinds of wireless systems. In this paper, two such techniques, MDS method and MLE that are recently proposed for collaborative location estimation, are studied in detail. From the theoretical formulation of the RSS-based location estimation problem, it is seen that MLE is more appropriate than MDS method. However, from simulation studies of both algorithms, which are iterative in nature, it is found that MLE is more sensitive to initial estimate than MDS method. Therefore, in this paper we propose to integrate these two techniques in series so that an estimate is first obtained using MDS method by taking advantage of its better convergence property, then MLE is employed to fine-tune the solution of MDS method to remove modeling errors that are inherent in MDS method. Through extensive simulations it is demonstrated that the new integrated method, named MDS-MLE, consistently outperforms both MDS method and MLE in various simulation scenarios. In this paper, we also address many important issues in collaborative localization, including effects of sensor node density, reference node density, and different deployment strategies of reference nodes.