RF-DNA Fingerprinting for the Detection of Malicious USB Devices

USB devices are considered to be a leading threat vector and a source for malicious attacks because they are used as a tool for updating and maintaining system configurations. The purpose of this research study is to evaluate the application of the Radio Frequency-Distinct Native Attributes (RF-DNA) fingerprinting technique on USB devices to discriminate between regular and malicious USB flash drives. We conducted an experiment by using the Unintentional Radiated Emissions (URE) produced by the normal function of the USB electronic components. We collected unintentional magnetic emissions from five USB devices under test used to develop Radio Frequency Distinct Native Attribute (RF-DNA) fingerprints that are used to classify and discriminate between different USB devices. We further used the Random Forest statistical learning algorithm for device classification and achieved 100\\% classification accuracy. Our results show the effectiveness and efficiency of the techniques in discriminating between regular and malicious USB devices. We conclude that our proposed technique can be used for preventing the use of malicious USB devices in sensitive organizations. Future work should study the effect of different noise levels on the classification accuracy and explore intra-manufacturer classification of USB devices.