Improved Radiometric Identification of Wireless Devices Using MIMO Transmission

Authenticating wireless devices based on features of their transmitted waveform has become a topic of considerable interest. Recent work in this arena has shown that examination of features in a modulated waveform can lead to highly precise identification of 802.11 devices. This paper experimentally demonstrates improved identification accuracy and reduced sensitivity to identification parameters when such techniques are applied to the multiple transmitters within multiple-input multiple-output devices. It further studies an information theoretic technique for determining which radiometric features are most effective in discriminating devices and analyzes the identification performance as a function of the number of features used. Finally, because some radiometric features are sensitive to environmental conditions and, therefore, drift with time, the analysis explores the impact of radiometric feature drift both on feature prioritization and on identification performance.