Research Files

Overview about files that came out of our research.

Models used for known-plaintext attacks (KPA) against M-209

Jupyter notebooks and files from various authors can be found in the GitHub repository.

File nameDescriptionDateSize
models.zip

This archive contains neural network models designed for the decryption of wheel 1 pins on the M-209. It includes models that process sequences of 52, 104, and 200 characters, each capable of taking pseudorandom displacement values generated by the machine and predicting the corresponding bit of the key.

2024-03-274.33 GB

M-209 cryptanalysis data

File nameDescriptionDateSize
1_keys_train.zip

A collection of keys generated according to the 1944 technical manual's specifications. This set underpins the creation of the training dataset.

2024-03-27537.6 MB
1_keys_test.zip

Consists of test keys, organized according to their cryptographic strength, which is determined by the number of non-shared and overlapping lugs.

2024-03-27116.2 MB
2_ciphertexts_train.zip

Contains JSON files detailing keys, keystreams, and ciphertexts derived from the 1_keys_train.zip keys, serving as the training dataset for the models.

2024-03-2714.55 GB
2_ciphertexts_test.zip

Similar to its training counterpart but generated using the test keys from 1_keys_test.zip, these files are used to test the models and analyse their performance based on the strength of the keys.

2024-03-2776.1 MB
3_data_npy_train.zip

Provides normalized data arrays that act as inputs and targets for models training.

2024-03-276.13 GB
3_data_npy_test.zip

Contains normalized data arrays that function as inputs and outputs during the testing phase of the models.

2024-03-2733.2 MB

Homophonic test ciphertexts

File nameDescriptionDateSize
Generated-Gutenberg-English-Homophonic-Ciphertexts.zip

English homophonic ciphertexts specifically created for testing and evaluating cryptanalysis algorithms. The ciphertexts were generated using the English Gutenberg Library. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1074.7 MB
Generated-Gutenberg-French-Homophonic-Ciphertexts.zip

French homophonic ciphertexts specifically created for testing and evaluating cryptanalysis algorithms. The ciphertexts were generated using the French Gutenberg Library. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1080.7 MB
Generated-Gutenberg-German-Homophonic-Ciphertexts.zip

German homophonic ciphertexts specifically created for testing and evaluating cryptanalysis algorithms. The ciphertexts were generated using the German Gutenberg Library. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1075.1 MB
Generated-Gutenberg-Hungarian-Homophonic-Ciphertexts.zip

Hungarian homophonic ciphertexts specifically created for testing and evaluating cryptanalysis algorithms. The ciphertexts were generated using the Hungarian Gutenberg Library. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1081.1 MB
Generated-Historic-English-1500-1800-Homophonic-Ciphertexts.zip

Historic English homophonic ciphertexts created for testing and evaluating cryptanalysis algorithms, featuring texts from 1500 to 1800. The ciphertexts were generated using the HistCorp dataset. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1974.4 MB
Generated-Historic-French-Homophonic-Ciphertexts.zip

Historic French homophonic ciphertexts created for testing and evaluating cryptanalysis algorithms, featuring texts from 1500 to 1800. The ciphertexts were generated using the HistCorp dataset. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1981.8 MB
Generated-Historic-German-1500-1800-Homophonic-Ciphertexts.zip

Historic German homophonic ciphertexts created for testing and evaluating cryptanalysis algorithms, featuring texts from 1500 to 1800. The ciphertexts were generated using the HistCorp dataset. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1974.9 MB
Generated-Historic-Hungarian-Homophonic-Ciphertexts.zip

Historic Hungarian homophonic ciphertexts created for testing and evaluating cryptanalysis algorithms. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-05-1976.2 MB
Generated-Historic-Italian-1500-1800-Homophonic-Ciphertexts.zip

Historic Italian homophonic ciphertexts created for testing and evaluating cryptanalysis algorithms, featuring texts from 1500 to 1800. The ciphertexts were generated using the HistCorp dataset. Homophonicity level (even) ranges between two and five. Also incorporates language-based homophonic ciphertexts (uneven).

2024-07-1471.3 MB

Historical language models for cryptanalysis

File nameDescriptionDateSize
Historic_English_Language_Models.zip

A collection of historic English language models for cryptanalysis. These models were trained on texts dating from the 16th to the 19th century.

2024-03-1350.6 MB
Historic_German_Language_Models.zip

A collection of historic German language models for cryptanalysis. These models were trained on texts dating from the 16th to the 19th century.

2024-03-15117 MB