Revealing IoT Cryptographic Settings through Electromagnetic Side-Channel Analysis

Authors: Zunaidi, Muhammad Rusyaidi; Sayakkara, Asanka and Scanlon, Mark

Publication Date: April 2024

Publication Name: Electronics

Abstract:

The advancement of cryptographic systems presents both opportunities and challenges in the realm of digital forensics. In an era where the security of digital information is crucial, the ability to non-invasively detect and analyse cryptographic configurations becomes significant. As cryptographic algorithms become more robust with longer key lengths,they provide higher levels of security. However, non-invasive side channels, specifically through electromagnetic (EM) emanations, can expose confidential cryptographic details, thus presenting a novel solution to the pressing forensic challenge. This research delves into the capabilities of EM Side-Channel Analysis (EM-SCA) specifically focused on detecting both cryptographic key lengths and the algorithms employed, utilising a machine learning-based approach, which can be instrumental for digital forensic experts during their investigations. Data collection was carried out on an Arduino Nano board, which executed the Advanced Encryption Standard (AES) and Elliptic Curve Cryptography (ECC) algorithms. Specifically, the board was tested with key lengths of 128, 192, and 256 for AES and 160, 192, and 256 for ECC. A HackRF One software-defined radio (SDR) facilitated the capture of EM emissions. A pipeline was implemented to process raw EM data, extract frequency-domain features, and bucket this information for dimensionality reduction, enhancing its applicability for Machine Learning (ML). ML models, such as Logistic Regression, Random Forest, XGBoost, LightGBM and Support Vector Machine (SVM), were trained on this processed dataset to differentiate between key lengths. Training multiple ML models on this specific dataset yielded varying degrees of accuracy in differentiating between key lengths. In a combined data examination of AES and ECC, the SVM model emerged with an accuracy of 94.55%. When individually assessed on AES and ECC data, Logistic Regression performed best accuracies of 98.47% and 98.76%, respectively. SVM once again demonstrated its ability in binary classification tasks between AES and ECC, obtaining an accuracy of 95.97%. This study contributes significantly to enhancing digital forensic capabilities in encrypted data investigation, offering a methodological advancement for non-invasively uncovering cryptographic settings in IoT devices.

Download:

Download Paper as PDF

BibTeX Entry:

@article{zunaidi2024IoTCrypto,
author={Zunaidi, Muhammad Rusyaidi and Sayakkara, Asanka and Scanlon, Mark},
title="{Revealing IoT Cryptographic Settings through Electromagnetic Side-Channel Analysis}",
journal="{Electronics}",
year=2024,
pages = {},
volume = {},
month=04,
issn = {2079-9292},
abstract={The advancement of cryptographic systems presents both opportunities and challenges in the realm of digital forensics. In an era where the security of digital information is crucial, the ability to non-invasively detect and analyse cryptographic configurations becomes significant. As cryptographic algorithms become more robust with longer key lengths,they provide higher levels of security. However, non-invasive side channels, specifically through electromagnetic (EM) emanations, can expose confidential cryptographic details, thus presenting a novel solution to the pressing forensic challenge. This research delves into the capabilities of EM Side-Channel Analysis (EM-SCA) specifically focused on detecting both cryptographic key lengths and the algorithms employed, utilising a machine learning-based approach, which can be instrumental for digital forensic experts during their investigations. Data collection was carried out on an Arduino Nano board, which executed the Advanced Encryption Standard (AES) and Elliptic Curve Cryptography (ECC) algorithms. Specifically, the board was tested with key lengths of 128, 192, and 256 for AES and 160, 192, and 256 for ECC. A HackRF One software-defined radio (SDR) facilitated the capture of EM emissions. A pipeline was implemented to process raw EM data, extract frequency-domain features, and bucket this information for dimensionality reduction, enhancing its applicability for Machine Learning (ML). ML models, such as Logistic Regression, Random Forest, XGBoost, LightGBM and Support Vector Machine (SVM), were trained on this processed dataset to differentiate between key lengths. Training multiple ML models on this specific dataset yielded varying degrees of accuracy in differentiating between key lengths. In a combined data examination of AES and ECC, the SVM model emerged with an accuracy of 94.55%. When individually assessed on AES and ECC data, Logistic Regression performed best accuracies of 98.47% and 98.76%, respectively. SVM once again demonstrated its ability in binary classification tasks between AES and ECC, obtaining an accuracy of 95.97%. This study contributes significantly to enhancing digital forensic capabilities in encrypted data investigation, offering a methodological advancement for non-invasively uncovering cryptographic settings in IoT devices.}
}