Damilola Adesina, Prairie View A&M University

Internet of Things Devices Fingerprinting using Deep Learning

Abstract: Radio Frequency (RF) fingerprinting could be used as a physical layer authentication method to distinguish legitimate wireless devices from adversarial ones. In this work, we present a wireless device identification platform to improve Internet of Things (IoT) security using Deep Learning techniques. Deep Learning is a promising method for obtaining the characteristics of the different RF devices through learning from their RF data. Specifically, three different deep learning models, namely Deep Neural Network (DNN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) are considered here to identify wireless devices and to distinguish among wireless devices from the same manufacturer.  As a case study, large data sets of RF traces from six “identical” ZigBee devices are collected using a Universal Software Radio Peripheral (USRP) based test bed. We captured RF data across a wide range of Signal-to-Noise Ratio (SNR) levels to guarantee the resilience of our proposed models to a variety of wireless channel conditions in practical scenarios. Experimental results demonstrate high accuracy of Deep Learning methods for wireless device identification that could potentially enhance IoT security.

Presentation Author(s):
Damilola Adesina*

Judging Forms Official judges only