Asia Othman Aljahdali, Maria Jawah, Jana Bakhalqi and Talah Fairaq
Adv. Artif. Intell. Mach. Learn., 5 (4):4765-4796
1. Asia Othman Aljahdali: University of Jeddah
2. Maria Jawah: University of Jeddah
3. Jana Bakhalqi: University of Jeddah
4. Talah Fairaq: University of Jeddah
DOI: 10.54364/AAIML.2025.54264
Article History: Received on: 21-Sep-25, Accepted on: 19-Dec-25, Published on: 26-Dec-25
Corresponding Author: Asia Othman Aljahdali
Email: aoaljahdali@uj.edu.sa
Citation: Asia Othman Aljahdali, et al. Detecting Wireless Relay Attacks in NFC Using Deep-Learning. Advances in Artificial Intelligence and Machine Learning. 2025;5(4):264. https://dx.doi.org/10.54364/AAIML.2025.54264
This study explores the application of deep learning (DL) to enhance security in Near Field Communication (NFC) technology, which is widely used in secure access
control and contactless payments. As NFC usage grows, concerns have emerged about security vulnerabilities, particularly relay attacks, where attackers relay sig-
nals without breaking encryption or other protective measures. Previous research focused on ambient-based, distance-bounding protocols and deep learning with RF
fingerprinting via Wi-Fi to mitigate such threats. This study will improve detecting of NFC relay attacks using RF fingerprints and deep learning via Bluetooth.
SDR++ software and a HackRF device were used to gather 2,400 NFC signal samples. 1,200 samples are allocated to the Normal category, while another 1,200
samples are assigned to the Relay Attack category. The dataset is trained, validated, and tested using the 2D-CNN model. The model’s test set accuracy was
88%, with 0.85 precision and 0.91 recall for the ”Normal” class and 0.90 precision and 0.84 recall for the ”Attack” class. Both classes’ F1-scores were approximately
0.88, showing performance that was balanced between sensitivity and precision.