KiSeok Bae

A new fault cryptanalysis on montgomery ladder exponentiation algorithm

The Montgomery ladder exponentiation algorithm is recognized as a very efficient countermeasure against Simple Power Analysis and C Safe-Error Attacks on RSA or elliptic curve cryptosystem. In this paper, we demonstrate the vulnerability of the Montgomery ladder algorithm to fault analysis attack when an error is injected during its operation in an embedded cryptographic chip. After injecting an error, we measure the power traced and compare it with an original correct trace. As a result, we can derive the secret key of the public-key cryptosystems such as RSA by computing the correlation coefficients of two power traces for correct and faulty cryptographic operations with same input.

Enhanced Stream Cipher Rabbit Secure Against Power Analysis Attack

Recently, stream cipher Rabbit was selected for the final eSTREAM portfolio organized by EU ECRYPT and as one of algorithm in part of ISO/IEC 18033-4 Stream Ciphers on ISO Security Standardization. However, a feasibility of practical power analysis attack to algorithm in experiment was introduced. Therefore, we propose appropriate methods such as random masking and hiding schemes to secure against power analysis attack on stream cipher Rabbit. We implement the proposed method with increment of 24% operating time and 12.3% memory requirements due to maintaining a high-speed performance. We use a 8-bit RISC AVR microprocessor (ATmegal128L chip) to implement our method for practical experiments, and verify that stream cipher Rabbit with our method is secure against power analysis attack.