Identification of electronic circuit boards through electromagnetic emission signals by using deep learning

Abstract

The ability to accurately distinguish between electronic devices is essential for security, especially in critical infrastructures. To prevent risks such as device spoofing and unauthorized access, reliable identification methods based on intrinsic physical signals are needed. In this study, near-field magnetic emissions from two Raspberry Pi 3B devices with identical hardware and software were captured using a near-field probe under different ADC configurations. The recorded signals were used to train a deep learning model to learn device-specific electromagnetic signatures. Performance was evaluated over a long measurement period with a large dataset to test stability and robustness. The effects of sampling parameters and controlled noise conditions were also analyzed. Results show that near-field electromagnetic emissions contain distinctive features suitable for device identification, and proper signal acquisition settings are critical for reliable classification.