Tracking in dynamic anchorless wireless networks based on manifold flattening

Abstract

We present a novel methodology for anchorless tracking of nodes in mobile ad hoc networks, where only range measurements between nodes that are within the communication range of each other are known. We assume that the network consists of both fixed and mobile nodes, and that the location of none of the nodes is known. Tracking is performed relative to the rigid structure provided by the fixed nodes. Our methodology extends our recently proposed Manifold Flattening (MF) algorithm to dynamic scenarios. By exploiting temporal correlation several of the inherent problems typical to static localization can be overcome. Temporal prediction of unknown range values can help the basic MF algorithm complete the Euclidean distance matrix in situations of poor network connectivity, and can also help eliminate both outliers in measurements or bad range estimations associated with the simplex flattening process.