Improving Localization in Sensor Networks by Leveraging Inter-node Ranking of Received Signal Strength and Clustering

摘要

Received Signal Strength (RSS) is notoriously inaccurate when used for range estimates in localization applications of wireless sensor networks. This paper demonstrates that better localization accuracies can be achieved by making the inter-node distance conform to the ranks of inter-node RSS, instead of the actual received signal strength when using the common Multi-Dimensional Scaling (MDS) localization approach. In order to further improve on the non-metric MDS localization results structural clustering constraints to the nodes were included in the optimization objective function. Experimental and simulation results show that the more clustering constraints that are included the better the localization results.