| Autor: |
Jonathan Martínez Padilla, Uwe Meyer-Baese, Simon Foo |
| Jazyk: |
angličtina |
| Rok vydání: |
2018 |
| Předmět: |
|
| Zdroj: |
EURASIP Journal on Information Security, Vol 2018, Iss 1, Pp 1-16 (2018) |
| Druh dokumentu: |
article |
| ISSN: |
2510-523X |
| DOI: |
10.1186/s13635-018-0073-z |
| Popis: |
Abstract In this work, side-channel attacks (SCAs) are considered as a security metric for the implementation of hybrid cryptosystems utilizing the neural network-based Tree Parity Re-Keying Machines (TPM). A virtual study is presented within the MATLAB environment that explores various scenarios in which the TPM may be compromised. Performance metrics are evaluated to model possible embedded system implementations. A new algorithm is proposed and coined as Man-in-the-Middle Power Analysis (MIMPA) as a means to copy the TPM’s generated keys. It is shown how the algorithm can identify vulnerabilities in the physical device in which the cryptosystem is implemented by using its power emissions. Finally, a machine learning approach is used to identify the capabilities of neural networks to recognize properties of keys produced in the TPM as they are transferred to an encryption algorithm. The results show that physical exploits of TPM implementations in embedded systems can be identified and accounted for before a final release. The experiments and data acquisition is demonstrated with an implementation of a TPM-AES hybrid cryptosystem in an AVR microcontroller. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
Full text from SpringerLink