Effect of Pivot Nodes Selection Schemes on Self-Localization Performance in a Mobile Sensor Network

To make a mobile sensor network allow better coverage of its sensable area, each mobile sensor node needs to recognize the locations of other nodes in a set of coordinates autonomously generated and shared in the network. To uniquely determine the location of a node on a plane, at least three pivot nodes should be selected from all nodes in the network and the scheme of selecting pivot nodes affects the accuracy of the estimated locations of the nodes. This paper discusses three distributed schemes of selecting pivot nodes such as far, random and near for a self-localization problem in a mobile sensor network, and theoretically analyzes the Cramer-Rao lower-bound on their location estimation error variance. Finally, the theoretically predicted result is confirmed by computer simulations and experiments.