On m-Permutation Protection Scheme against Modification Attack

Abstract

The EEPROM modification attack was first described by Anderson and Kuhn in [3]. This simple and low-cost attack is very efficient against tamperproof devices carrying a secret key stored in EEPROM. Soon after the attack has been published, we proposed a protection scheme using cascaded m-permutations of hidden wires [8]. This cascaded m-permutation protection scheme uses an (m×n)-bit encoding for an n-bit key and for which the best known attack will take at most O(n ^m) probes to compromise the permutations of the hidden wires. However, it is observed that if a particular card (instead of the whole batch of cards) is to be compromised, the complexity can be greatly reduced, and in the best cases, it can even be reduced to linear time complexity. In this paper, we demonstrate how it can be done, and propose a revised m-permutation scheme that would close the loop-hole. It is also proved that the probability of breaking the revised scheme will be 1/2ⁿ⁻¹ for a n-bit key.