Support Vector Machines for Non-linear Radio Fingerprint Recognition in Real-life Industrial Environments

A. Lewandowski, V. Köster, C. Wietfeld, S. Michaelis

Abstract:	Real-life industrial environments are challenging for today's radio-based localization systems, as the radio channel is highly affected by the circumstances of the environment. Therefore, this paper presents a solution for non-linear radio fingerprint recognition based on Support Vector Machines in order to profit from the disperse circumstances of such an environment. After specifying the sophisticated modeling approach, a detailed description about the used algorithms is given, before a comparison between different parameterizations and physical circumstances highlights the performance of such a system in terms of accuracy and reliability.
Published in:	Proceedings of the 2011 International Technical Meeting of The Institute of Navigation January 24 - 26, 2011 Catamaran Resort Hotel San Diego, CA
Pages:	628 - 634
Cite this article:	Lewandowski, A., Köster, V., Wietfeld, C., Michaelis, S., "Support Vector Machines for Non-linear Radio Fingerprint Recognition in Real-life Industrial Environments," Proceedings of the 2011 International Technical Meeting of The Institute of Navigation, San Diego, CA, January 2011, pp. 628-634.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In