Radiometric identification of LTE transmitters
Frédéric Demers, Marc St‐Hilaire · 2013
This paper demonstrates that highly accurate radiometric identification of Long Term Evolution (LTE) transmitters is possible using commercial off-the-shelf hardware and support vector machines (SVM). The identification is based on unique modulation characteristics exhibited by the transmitters, resulting from minute imperfections introduced during radio hardware manufacturing. In these experiments, the Agilent Vector Signal Analysis (VSA) software and the Agilent PXA spectrum analyzer are used to extract radiometric properties from several LTE base stations, known as evolved Node B (eNB). The open-source SVM library libsvm performs the classification using 13 feature coefficients extracted by the VSA. When SVM parameters are optimized using grid search, and the training bin contains no less than 45 vectors, re-identification is shown to be in excess of 98%.