Deep Learning Classification of Motherboard Components by Leveraging EM Side-Channel Signals

Jorgensen, Erik J.; Werner, Frank T.; Prvulovic, Milos; Zajić, Alenka

doi:10.1007/s41635-021-00116-2

Cited by 7 publications

(1 citation statement)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Studies in [7][8][9][10][11] showed that methods based on spectrum analysis of decoupled electromagnetic (EM) emissions are promising to find malware in control system processors. This monitoring may be carried out remotely due to its decoupled construction.…”

Section: Introductionmentioning

confidence: 99%

Hybrid Quantum Artificial Intelligence Electromagnetic Spectrum Analysis Framework for Transportation System Security

Vaidyan

Rimal

2023

Preprint

View full text Add to dashboard Cite

The traffic signal control (TSC) system is faced with both opportunities and challenges as a consequence of connected vehicle (CV) technology. Although implementing CV technology might considerably enhance safety and mobility performance, the connection between vehicles and transportation infrastructure may raise the dangers of cyber-attacks. Studies on cybersecurity in TSC systems have been undertaken in recent years. However, a safe framework for air-gapped malware analysis is still lacking. Our study aims to close this research gap by presenting a thorough electromagnetic analysis approach to address the cybersecurity issue facing the TSC in the CV environment. In this technique, a hybrid deep-learning architecture based on classical-quantum transfer learning models is constructed to study electromagnetic (EM) spectra. We assess the effect of adversarial attacks on TSC systems using these hybrid models and distinguish attacks from normal operations of the controller. A neural network trained on a dataset to collect pertinent characteristics from a high-dimensional dataset of Electromagnetic (EM) trace-based TSC attack vectors form the basis of hybrid models. The outcome of the standard deep learning process is then examined by the quantum layer. A number of quantum gates make up the quantum layer, which can enable a number of quantum mechanical activities, including superposition and entanglement. The most conceivable and feasible attack approach in transport signal controllers has been found to be data spoofing, and the potential of the proposed air-gapped electromagnetic Hybrid classical quantum monitoring framework in sensor fusion and interrupts is explored in light of these findings.

show abstract