Topology Insensitive Location Determination Using Independent Estimates Through Semi-Directional Antennas
Date of this Version11-1-2006
This document has been peer-reviewed.
We demonstrate the effect of using multiple estimations from independent single wireless motes in order to decrease network topology dependence on location estimation in a wireless sensor network. A method of determining the location of a target by using multiple compact semi-directional antennas is shown to give an independent estimate of location from each sensor mote in a network, each estimate not relying on the data from neighboring motes as in the case of traditional triangulation. We begin by demonstrating a method of using angular diversity through multiple semi-directional antennas in order to ascertain the location of a target. The estimation of both range and angle is demonstrated in the presence of a noisy and/or faded channel. An efficient and fast algorithm on a wireless sensor mote is presented through a Taylor series expansion of the simulated antenna pattern. Furthermore, using the results from the location estimation from a single node, location determination in a realistic network is explored through both theory and simulation. The results indicate that our proposed algorithm depends significantly less on the topology (spatial arrangement) of the anchor nodes. While network planning for a variety of topologies of anchor nodes is shown to be necessary when using triangulation, our proposed algorithm is insensitive to the deployments of the anchor nodes. A testbed was created in order to experimentally demonstrate that the predictions are accurate even in triangulation-adverse topologies. The experimental testbed shows that a linear arrangement of closely spaced sensors can reduce the location error to one-fourth of the location error using triangulation.