Research on Grain-128's cryptosystem recognition

As an important aspect of distinguishing attack, cryptosystem recognition is the foundation of cryptanalysis. We mainly focused on the recognition of Grain-128 between other 11 cryptosystems. Firstly, we extracted 25 features of ciphertexts, then we constructed cryptosystem recognition classifier based on random forest algorithm. The recognition experiments between Grain-128 and other 11 cryptosystems were implemented. The results of experiments show that, in the condition of known ciphertext, Grain-128 can be effectively identified from other 11 cryptosystems, the performance of randomness test based features are better than other existed features with its accuracy of cryptosystem recognition average over 10%. With maintaining the performance of features, some features’ dimension reductions are completed and features’ data utilities are improved by t-SNE algorithm. Keywords—cryptosystem recognition; block cipher; randomness test; feature extraction; random forest