An evolutionary computation attack on one-round TEA

In this work, one-round Tiny Encryption Algorithm (TEA) is attacked with an Evolutionary Computation method inspired by a combination of Genetic Algorithm (GA) and Harmony Search (HS). The system presented evaluates and evolves a population of candidate keys and compares paintext-ciphertext pairs of the known key against said population. We verify that randomly generated keys are the hardest to derive. Keys composed of words containing all on-bits are more difficult to break than keys composed of words containing all off-bits. Keys which have repeated words are easiest to derive. Finally, the present EC strategy is capable of deriving degenerate keys; this is most evident when keys are front loaded so that the first byte of each word has the highest density of on-bits.