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

摘要

Real-life industrial environments are challenging forrntoday's radio-based localization systems, as the radiornchannel is highly affected by the circumstances of thernenvironment. Therefore, this paper presents a solution forrnnon-linear radio fingerprint recognition based on SupportrnVector Machines in order to profit from the disperserncircumstances of such an environment. After specifyingrnthe sophisticated modeling approach, a detailedrndescription about the used algorithms is given, before arncomparison between different parameterizations andrnphysical circumstances highlights the performance ofrnsuch a system in terms of accuracy and reliability.