Thunder

Abstract

Localization for outdoor wireless sensor networks has been a challenge for real applications. Although many solutions have been proposed, few of them can be used in real applications because of their high cost, low accuracy or infeasibility due to practical issues. In this paper, we propose a practical acoustic localization scheme called Thunder. Thunder employs an asymmetric architecture and shifts most of the complexities and hardware requirements from each node to a single powerful centralized device. The solution is efficient, and requires virtually zero cost in terms of extra per node hardware and innetwork communication. This paper also presents an efficient scheduling algorithm called Equilateral Triangle Scheduling to schedule Thunder for very large sensor networks and a resilient algorithm called Adaptive Fuzzy Clustering to provide robust localization without sacrificing efficiency in the presence of a high percentage of large ranging errors. To validate and evaluate Thunder, we built an experimental localization system based on the Mica2 platform, which achieved localization errors of about 1 meter in medium scale localization experiments.