Fault attacks and countermeasures for elliptic curve cryptosystems

Medos, S 2009, Fault attacks and countermeasures for elliptic curve cryptosystems, Doctor of Philosophy (PhD), Mathematical and Geospatial Sciences, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Medos.pdf Thesis Click to show the corresponding preview/stream application/pdf;... 695.07KB
Title Fault attacks and countermeasures for elliptic curve cryptosystems
Author(s) Medos, S
Year 2009
Abstract In this thesis we have developed a new algorithmic countermeasures that protect elliptic curve computation by protecting computation of the finite binary extension field, against fault attacks. Firstly, we have proposed schemes, i.e., a Chinese Remainder Theorem based fault tolerant computation in finite field for use in ECCs, as well as Lagrange Interpolation based fault tolerant computation. Our approach is based on the error correcting codes, i.e., redundant residue polynomial codes and the use of first original approach of Reed-Solomon codes. Computation of the field elements is decomposed into parallel, mutually independent, modular/identical channels, so that in case of faults at one channel, errors will not distribute to other channels. Based on these schemes we have developed new algorithms, namely fault tolerant residue representation modular multiplication algorithm and fault tolerant Lagrange representation modular multiplication algorithm, which are immune against error propagation under the fault models that we propose: Random Fault Model, Arbitrary Fault Model, and Single Bit Fault Model. These algorithms provide fault tolerant computation in GF (2k) for use in ECCs. Our new developed algorithms where inputs, i.e., field elements, are represented by the redundant residue representation/ redundant lagrange representation enables us to overcome the problem if during computation one, or both coordinates x, y GF (2k) of the point P E/GF (2k) /Fk are corrupted. We assume that during each run of an attacked algorithm, in one single attack, an adversary can apply any of the proposed fault models, i.e., either Random Fault Model, or Arbitrary Fault Model, or Single Bit Fault Model. In this way more channels can be targeted, i.e., different fault models can be used on different channels. Also, our proposed algorithms can have masked errors and will not be immune against attacks which can create those kind of errors, but it is a difficult problem to counter masked errors, since any anti-fault attack scheme will have some masked errors. Moreover, we have derived conditions that inflicted error needs to have in order to yield undetectable faulty point on non-supersingular elliptic curve over GF(2k). Our algorithmic countermeasures can be applied to any public key cryptosystem that performs computation over the finite field GF (2k).
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Mathematical and Geospatial Sciences
Keyword(s) Side channel attack
fault attack
elliptic curve cryptosystem
Versions
Version Filter Type
Access Statistics: 264 Abstract Views, 264 File Downloads  -  Detailed Statistics
Created: Wed, 12 Sep 2012, 14:17:19 EST by Brett Fenton
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us