Symbol Based Statistical RF Fingerprinting for Fake Base Station Identification

Abstract

The identification of fake base station (FBS) in a cellular network has become challenging with the development of various software defined radio platforms and mobile standards. This paper, therefore, presents robust statistical approach to detect unique non-linearities based hardware impairments of the transmitter. Employing the fact that power amplifier (PA) of a regular base station (RBS) is a costly and high precision device with a provision of sophisticated digital predistortion (DPD) hardware implementation and in contrary, this DPD based linearization effort is not spent in existing SDR platforms so PA of an SDR based FBS tends to violate the spectral mask and introduces large amplitude and phase errors in the transmitted signal compared to the RBS. At first, a second order symbol-based error vector magnitude (EVM) approach is triggered at the user equipment (UE) to measure the non-linearity induced by the PA of various SDR based FBS. Afterward, a higher fourth order moment i.e. kurtosis approach has been proposed along with the actual measurement results to determine the noise structuredness of the received signal at UE. The kurtosis on magnitude of extracted complex noise cloud is found to be a strong indicator to identify the FBS.