Variable elasticity spring-relaxation: improving the accuracy of sensor localization for unknown path loss exponent

Zhang, Q., Foh, C.H., Seet, B.C. and Fong, A.C.M. (2012) Variable elasticity spring-relaxation: improving the accuracy of sensor localization for unknown path loss exponent. Personal and Ubiquitous Computing, 16(7), pp. 929-941. (doi: 10.1007/s00779-011-0449-2)

Abstract

Wireless sensor network is a key enabling technology for Ambient Intelligence, where location information is crucial for many applications. RSS-based ranging localization takes advantage of its low cost and low complexity, but it has an infeasible assumption of an accurate path loss exponent of the physical environment. In this paper, we study the impact of path loss exponent accuracy on the localization accuracy. We formulate the relationship between the path loss exponent estimate and localization error, and found the localization error of exponential order which we call the error magnification effect. By our in-depth investigation, we propose a passive and an active measures to suppress the error magnification effect, where the passive measure stabilizes the localization error of the spring-relaxation algorithm (SR), and the active measure introduces variable elasticity into the SR algorithm to cancel off the exponential ranging error. The combination of both measures forms our localization solution called variable elasticity spring-relaxation (VE-SR) localization. We conduct extensive simulation experiments to show the effectiveness of VE-SR in suppressing the error magnification effect in various experiment setup. For a wide variety of physical environments, VE-SR offers location estimation with an average accuracy of no more than 10% of transmission range.

• Item Type: Articles
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Fong, Dr Alvis Cheuk Min
• Authors: Zhang, Q., Foh, C.H., Seet, B.C., and Fong, A.C.M.
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
• College/School: College of Science and Engineering > School of Computing Science
• Journal Name: Personal and Ubiquitous Computing
• Publisher: Springer Verlag
• ISSN: 1617-4909
• ISSN (Online): 1617-4917