leehanbo / advanced-far-field-em-side-channel-attack-onaes-based-on-deep-learning Goto Github PK
View Code? Open in Web Editor NEWThis project forked from nymeriawang/advanced-far-field-em-side-channel-attack-onaes-based-on-deep-learning
The reporitory contains scripts and data used in the paper <Advanced Far Field EM Side-Channel Attack on AES Based on Deep Learning>. This paper presents a deep learning-based side-channel attack on AES-128 using far field electromagnetic (EM) emissions as a side channel. Unlike power or near filed EM analysis, far field EM attacks do not require a close physical proximity to the device under attack. Our neural networks are trained on traces captured from five different Bluetooth devices and tested on five other Bluetooth devices. The training set is composed as a combination of ``clean'' traces, captured through a coaxial cable, and traces captured on a distance to device. Our best model can recover the key from less than 300 traces captured in an office environment at 15 m distance to device without repeating each encryption more than once.