DC-NFC: A Custom Deep Learning Framework for Security and Privacy in NFC-Enabled IoT

NFC has emerged as a critical technology in IoET ecosystems, facilitating seamless data exchange in proximity-based systems. However, the security and privacy challenges associated with NFC-enabled IoT devices remain significant, exposing them to various threats such as eavesdropping, relay attacks, and spoofing. This paper introduces DC-NFC, a novel deep learning framework designed to enhance the security and privacy of NFC communications within IoT environments. The proposed framework integrates three innovative components: the CE for capturing intricate temporal and spatial patterns, the PML for enforcing end-to-end privacy constraints, and the ATF module for real-time threat detection and dynamic model adaptation. Comprehensive experiments were conducted on four benchmark datasets-UNSW-NB15, Bot-IoT, TON-IoT Telemetry, and Edge-IIoTset. The results of the proposed approach demonstrate significant improvements in security metrics across all datasets, with accuracy enhancements up to 95% on UNSW-NB15, and consistent F1-scores above 0.90, underscoring the framework's robustness in enhancing NFC security and privacy in diverse IoT environments. The simulation results highlight the framework's real-time processing capabilities, achieving low latency of 20.53 s for 1000 devices on the UNSW-NB15 dataset.