A Generic attack on CubeHash, a SHA-3 candidate

A secure cryptographic hashing function should be resistant to three different scenarios: First, a cryptographic hashing function must be preimage resistant, that is, it should be infeasible for an attacker to construct a message such that it produces a known hash output value. Second, a cryptographic hashing function must be second preimage resistant, or it should be infeasible for an attacker to construct a message such that it has the same hash output value as another known message. Third, a cryptographic hashing function must be collision resistant, which means that it should be infeasible for an attacker to find any two different messages such that their hash output values are the same. The current Secure Hash Algorithm (SHA) family, namely SHA-1 and SHA-2, were designed by the National Security Agency (NSA) and published by the National Institute of Standards and Technology (NIST). Recent advances in cryptanalysis of hash functions have led to concerns about the collision resistance in the SHA family. To address these concerns, NIST has opened a public worldwide competition known as the SHA-3 competition to find the new hash function, which will become SHA-3. Each candidate hash function is scrutinized by the public, and candidates with found weaknesses are dropped from advancing to the next rounds of the competition. The goal is that the strongest hash function will emerge at the end of the competition, and this hash function will be free for everyone to use. This thesis implemented a generic attack against the collision resistance of small variants of one candidate in the SHA-3 competition, CubeHash. A unique hash-chaining approach was used to find the collisions, and the parallelization of several FPGAs lead to parallelization measurements and analysis to see if a linear speedup could be obtained