Given the ever increasing use of near field communication (NFC), the security of this system is becoming increasingly important. Recently, radio frequency (RF) fingerprinting, where the physical RF characteristics of a communication device are used as a means to provide guarantees of authenticity and security, has received serious consideration due to the uniqueness of these characteristics, making cloning difficult. In this article, we discuss the feasibility of RF fingerprinting assisted by deep learning for use in identifying NFC tags. To this end, we implement a hardware testbed with an off-the-shelf NFC reader and software defined radio. An RF signal corresponding to one-bit transmission from the NFC tag is used to extract the RF characteristics, which enables rapid identification. Three different types of deep neural network are used, namely fully connected layer-based neural network' convolutional neural network, and recurrent neural network. By exper-iment, we confirm that deep-learning-based algorithms can uniquely distinguish 50 NFC tags with up to 96.16 percent accuracy. We also discuss some of the key technical challenges involved in the use of deep-learning-based RF fingerprinting for NFC.
|Number of pages||6|
|Journal||IEEE Communications Magazine|
|Publication status||Published - 2021 May|
Bibliographical noteFunding Information:
This research was supported in part by the Affiliated Institute of the Electronics and Telecommunications Research Institute (ETRI), in part by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2018R1D-1A1B07040796), and in part by Korea National University of Transportation in 2020.
© 1979-2012 IEEE.
All Science Journal Classification (ASJC) codes
- Computer Science Applications
- Computer Networks and Communications
- Electrical and Electronic Engineering