Localization and sensing applications in the Pushpin Computer Network

Abstract

The utility and purpose of a node in a wireless sensor network is intimately tied to the physical space in which it is distributed. As such, it is advantageous under most circumstances for a sensor node to know its position. In this work, we present two systems for localizing a network of roughly 60 sensor nodes distributed over an area of 1-m2. One is based on a linear lateration technique, while the second approach utilizes non-linear optimization techniques, namely spectral graph drawing and mesh relaxation. In both cases, localization is accomplished by generating distance constraints based on ultrasound time-of-flight measurements to distinct, global sensor stimuli. These distance constraints alone are sufficient to achieve localization; no a priori knowledge of sensor node coordinates or the coordinates of the global sensor events are required. Using this technique, we have achieved a localization error of 2.30-cm and an error standard deviation of 2.36-cm.