Title :
Comparative Power Analysis of Modular Exponentiation Algorithms
Author :
Homma, Naofumi ; Miyamoto, Atsushi ; Aoki, Takafumi ; Satoh, Akashi ; Samir, Akhrouf
Author_Institution :
Grad. Sch. of Inf. Sci., Tohoku Univ., Sendai, Japan
fDate :
6/1/2010 12:00:00 AM
Abstract :
This paper proposes new chosen-message power-analysis attacks for public-key cryptosystems based on modular exponentiation, where specific input pairs are used to generate collisions between squaring operations at different locations in the two power traces. Unlike previous attacks of this kind, the new attack can be applied to all standard implementations of the exponentiation process, namely binary (left-to-right and right-to-left), m-ary, and sliding window methods. The proposed attack can also circumvent typical countermeasures, such as the Montgomery powering ladder and the double-add algorithm. The effectiveness of the attack is demonstrated in experiments with hardware and software implementations of RSA on an FPGA and a PowerPC processor, respectively. In addition to the new collision generation methods, a highly accurate waveform matching technique is introduced for detecting the collisions even when the recorded signals are noisy and there is a certain amount of clock jitter.
Keywords :
field programmable gate arrays; jitter; public key cryptography; FPGA; Montgomery powering ladder; PowerPC processor; binary method; chosen-message power-analysis attacks; clock jitter; collision generation methods; comparative power analysis; double-add algorithm; m-ary method; modular exponentiation algorithms; public-key cryptosystems; sliding window method; waveform matching technique; Algorithm design and analysis; Cathode ray tubes; Clocks; Energy consumption; Field programmable gate arrays; Hardware; Noise generators; Power generation; Public key cryptography; Signal generators; RSA; Side-channel attacks; modular exponentiation; power-analysis attacks; waveform matching.;
Journal_Title :
Computers, IEEE Transactions on
DOI :
10.1109/TC.2009.176