| Title | Detection of Trojan Based DoS Attacks on RSA Cryptosystem Using Hybrid Supervised Learning Models |
| Publication Type | Conference Paper |
| Year of Publication | 2020 |
| Authors | Gayatri, R, Gayatri, Yendamury |
| Conference Name | 2020 Third International Conference on Smart Systems and Inventive Technology (ICSSIT) |
| Date Published | aug |
| Keywords | chipwhisperer, composability, cryptography, cyber physical security, cyber physical systems, Data models, Hardware, hardware Trojan horses, Integrated circuit modeling, machine learning, machine learning algorithms, Power profile, pubcrawl, resilience, Resiliency, RSA, security, supply chain security, trojan horse detection, Trojan horses |
| Abstract | Privacy and security have become the most important aspects in any sphere of technology today from embedded systems to VLS I circuits. One such an attack compromising the privacy, security and trust of a networked control system by making them vulnerable to unauthorized access is the Hardware Trojan Horses. Even cryptographic algorithms whose purpose is to safeguard information are susceptible to these Trojan attacks. This paper discusses hybrid supervised machine learning models that predict with great accuracy whether the RSA asymmetric cryptosystem implemented in Atmel XMega microcontroller is Trojan-free (Golden) or Trojan-infected by analyzing the power profiles of the golden algorithm and trojan-infected algorithm. The power profiles are obtained using the ChipWhisperer Lite Board. The features selected from the power profiles are used to create datasets for the proposed hybrid models and train the proposed models using the 70/30 rule. The proposed hybrid models can be concluded that it has an accuracy of more than 88% irrespective of the Trojan types and size of the datasets. |
| DOI | 10.1109/ICSSIT48917.2020.9214116 |
| Citation Key | gayatri_detection_2020 |
Detection of Trojan Based DoS Attacks on RSA Cryptosystem Using Hybrid Supervised Learning Models