The leakage model as the dominant factor in correlation power analysis of AES-1298 on a CORTEX-M0
2026, vol.18 , no.4, pp. 115-126
Article [2026-04-11]
Cryptographic algorithms secure at the algorithmic level may remain vulnerable when executed on physical hardware. This paper characterises correlation power analysis against an unprotected AES-128 implementation on a low-cost ARM Cortex-M0 microcontroller, separating four factors commonly assumed to govern attack success: the leakage model, the trace count, signal preprocessing, and the choice between profiled and non-profiled attacks. Two datasets of 5 000 and 10 000 traces were captured on a ChipWhisperer-Nano and analysed with a custom implementation cross-validated against the ChipWhisperer Analyzer. The complete key was recovered from 55 traces. The Hamming weight model gave a mean peak correlation of 0.6985 against 0.0996 for Hamming distance, which resolved only 2 of 16 bytes; filtering altered correlation by 0.45%. Leakage model choice dominates every other factor examined.
side-channel analysis, correlation power analysis, leakage model, AES-128, embedded security
https://doi.org/10.59035/VYXE5307
Gabriela Nedyalkova, Magdalena Garvanova, Ivan Garvanov. The leakage model as the dominant factor in correlation power analysis of AES-1298 on a CORTEX-M0. International Journal on Information Technologies and Security, vol.18 , no.4, 2026, pp. 115-126. https://doi.org/10.59035/VYXE5307