The Design of Calamari : an Ad - hoc Localization System for Sensor Networks

The Design of Calamari: an Ad-hoc Localization System for Sensor Networks by Cameron Dean Whitehouse Master of Science in Electrical Engineering and Computer Science University of California at Berkeley Professor David Culler, Research Advisor This thesis describes the design of an ad-hoc localization system for sensor networks. We begin by formulating design principles from a concrete implementation of a sensor network application that requires localization. We show why several existing localization systems are unsatisfactory for this application domain and propose a new system called Calamari that is designed specifically for this domain’s goals and constraints. We motivate and explain the design decisions pertinent to Calamari’s ranging technology, calibration and localization algorithm. The ranging technologies ultimately used are extremely cheap and simple and have approximately 30% error after simple calibration. A new calibration procedure is designed that reduces this error to about 10%. A modificiation to APS, an existing distributed localization algorithm, is proposed that improves error by 20% to 30%. Preliminary experiments show that this ranging error with this algorithm is sufficient to give rise to position estimates that are on average 20% of the maximum distance between connected nodes in the network.