Support Vector Machine - Based Classification of Wireless Transceivers

The wireless device authentication based on the impairments of radio frequency front-end is a promising method how to increase the physical layer security of future wireless networks. This paper is demonstrating the use of one of the most-used machine learning methods - Support Vector Machines in such an application. Besides sketching the multi-class authentication on an example of data from a set of software defined radios, we also evaluate how is the used classifier sensitive to changes of working temperature during the learning and testing phases.