Shen, G, Zhang, J and Marshall, A
(2024)
Deep learning-enhanced radio frequency fingerprint identification.
PhD thesis, University of Liverpool.
|
Text
201457466_June2023.pdf - Author Accepted Manuscript Download (7MB) | Preview |
Abstract
Radio frequency fingerprint identification (RFFI) is a promising authentication technique for Internet of Things (IoT) devices. The wireless transmitter chain of an IoT device consists of analog components that inevitably deviate from nominal values during the manufacturing process. These imperfections slightly distort the transmitted waveform, and an RFFI system can be equipped at the receiver to uniquely identify the device by analyzing the characteristics of the received signal. The deep learning technique is frequently used in RFFI for its effective feature extraction capabilities. After sufficient training, the neural network is capable of predicting the device identity by analyzing the input signal. In this chapter, we first introduce the basic components of an RFFI system and then use LoRa as a case study to demonstrate implementation details. The channel-independent spectrogram is proposed as the neural network input to combat channel effects. Then we design CNN and LSTM networks to process the channel-independent spectrogram. Experiments are conducted involving five LoPy4 devices and a USRP N210 software-defined radio (SDR). The results demonstrate that the designed LoRa-RFFI system is resilient to location changes, and the classification accuracy is consistently above 90%.
| Item Type: | Thesis (PhD) |
|---|---|
| Divisions: | Faculty of Science and Engineering > School of Electrical Engineering, Electronics and Computer Science |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 19 Jun 2023 15:09 |
| Last Modified: | 15 Jun 2024 07:02 |
| DOI: | 10.17638/03170842 |
| Supervisors: |
|
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3170842 |
Altmetric
Altmetric
