Towards Automated Identification of Block Cipher Structures Using Machine Learning

Uroosa Kiran; Hammad Tanveer Butt; Zunera Jalil

doi:10.2478/ias-2026-0007

.blurhash-client-img { display: none !important; }

Towards Automated Identification of Block Cipher Structures Using Machine Learning

Journal of Information Assurance and Security

Volume 21 (2026): Issue 3 (July 2026)

By: Uroosa Kiran, Hammad Tanveer Butt and Zunera Jalil

Open Access

|Jul 2026

Figures & Tables

Framework for Cipher Structure Classification Using Machine Learning

Confusion Matrices of all ML Models (a; XGBoost, b; RF, c; NB, d; LR, e;KNN) for structure classification under S1 for 200KB.

Confusion Matrices of ML models (a; XGBoost, b; RF, c; NB, d; LR, e;KNN) for structure classification under S2 for 200KB.

Statistical relationships among ciphertext features

Feature Analysis of Classifiers for CBC and ECB modes (a) Fixed key (b) Random Kyes

Performance Analysis of Classifiers for CBC and ECB modes (a) Fixed key (b) Random Key

Model Statistics and Performance Distribution (CBC) (a) Fixed key (b) Random Key

NIST Test Pass Rates (%) over File Size for CBC mode (S1, S2)

NIST Test Pass Rates (%) over File Size for ECB mode (S3, S4)

Binary Classification Results for S1_

Metric	File Size (KB)	XGBoost	RF	NB	LR	KNN
Accuracy	10	0.93	0.93	0.72	0.77	0.79
	20	0.92	0.93	0.72	0.80	0.80
	50	0.92	0.92	0.72	0.80	0.80
	100	0.92	0.93	0.72	0.92	0.92
	200	0.92	0.93	0.72	0.92	0.92
Precision	10	0.93	0.96	0.92	0.80	0.78
	20	0.93	0.95	0.95	0.82	0.81
	50	0.93	0.97	0.99	0.85	0.83
	100	0.92	0.99	1.00	0.87	0.92
	200	0.92	0.98	1.00	0.86	0.92
Recall	10	0.93	0.93	0.52	0.76	0.90
	20	0.92	0.93	0.50	0.79	0.80
	50	0.91	0.92	0.52	0.79	0.80
	100	0.93	0.93	0.56	0.86	0.92
	200	0.93	0.93	0.47	0.87	0.93

Average Number of Words against File Sizes

File Size (kb)	Average Word Length
10	1,296
20	2,589
50	6,494
100	12,976
200	25,805

Experimental Scenarios

Scenario	Mode	Key	IV
S1	CBC	Fixed	Fixed
S2	CBC	Random	Fixed
S3	ECB	Fixed	Fixed
S4	ECB	Random	Fixed

Diverse Dataset for ML Models

Category	Classes	No. of Files	File Sizes (kb)	Ciphers	Modes	No. of Keys	Encrypted Files	Source/Dataset
Wikipedia Articles	67	134	10, 20, 50, 100, 200	16	CBC, ECB	10	134,000	HuggingFace / FineFineWeb
Twitter Data	1	100					100,000	Kaggle / Sentiment140
Mixed News	1	100					100,000	Kaggle / RealNews
Programming Codes	5	100					100,000	HuggingFace / XCodEval
Network Traffic	1	100					100,000	Kaggle / IPNetworkTrafficFlows

Binary Classification Results for S2

Metric	File Size (KB)	XGBoost	RF	NB	LR	KNN
Accuracy	10	0.92	0.93	0.70	0.72	0.78
	20	0.92	0.93	0.71	0.73	0.77
	50	0.92	0.93	0.71	0.73	0.78
	100	0.92	0.93	0.70	0.87	0.92
	200	0.92	0.94	0.70	0.88	0.93
Precision	10	0.92	0.93	0.75	0.73	0.77
	20	0.92	0.94	0.78	0.74	0.77
	50	0.92	0.97	0.82	0.74	0.78
	100	0.92	0.98	0.89	0.87	0.92
	200	0.92	0.99	0.96	0.88	0.93
Recall	10	0.92	0.93	0.61	0.85	0.79
	20	0.92	0.93	0.58	0.73	0.81
	50	0.92	0.93	0.53	0.73	0.78
	100	0.92	0.93	0.49	0.90	0.92
	200	0.92	0.93	0.49	0.87	0.93

Deep Learning (MLP) Results - CBC Mode

Scenario	File Size (KB)	Accuracy	Precision	Recall
S1	10	0.77	0.80	0.77
	20	0.80	0.85	0.78
	50	0.78	0.82	0.82
	100	0.88	0.88	0.90
	200	0.88	0.90	0.87
S2	10	0.76	1.00	0.77
	20	0.77	0.98	1.00
	50	0.77	1.00	0.79
	100	0.87	0.87	0.88
	200	0.88	0.88	0.87

Ciphers Used in Dataset Construction

Algorithm	Structure	Block Size	Modes	No. of Keys
DES	Feistel	64 bits	CBC, ECB	10
AES	SPN	128 bits
3DES	Feistel	64 bits
CAST	Feistel	64 bits
Blowfish	Feistel	64 bits
KASUMI	Feistel	64 bits
TWINE	Feistel	64 bits
SIMON	Feistel	128 bits
SM4	Feistel	128 bits
ARIA	SPN	128 bits
SERPENT	SPN	128 bits
PRESENT	SPN	64 bits
ARADI	SPN	128 bits
SKINNY	SPN	64 bits
KALYNA	SPN	256 bits
RIJNDAEL	SPN	128 bits

Binary Classification Results for S3

Metric	File Size (KB)	XGBoost	RF	NB	LR	KNN
Accuracy	10	0.98	0.95	0.82	0.91	0.94
	20	0.96	0.95	0.82	0.90	0.94
	50	0.97	0.95	0.83	0.92	0.95
	100	0.97	0.95	0.81	0.93	0.94
	200	0.99	0.97	0.73	0.95	0.96
Precision	10	0.98	0.99	1.00	0.95	0.94
	20	0.96	1.00	1.00	0.93	0.95
	50	0.98	1.00	1.00	0.96	0.95
	100	0.99	1.00	1.00	0.96	0.95
	200	1.00	1.00	1.00	0.99	0.96
Recall	10	1.00	0.95	0.75	0.90	0.94
	20	0.95	0.94	0.76	0.87	0.94
	50	0.96	0.94	0.89	0.89	0.94
	100	0.96	0.94	0.88	0.90	0.94
	200	0.98	0.96	0.87	0.93	0.96

Binary Classification Results for S4

Metric	File Size (KB)	XGBoost	RF	NB	LR	KNN
Accuracy	10	0.93	0.94	0.78	0.91	0.93
	20	0.92	0.94	0.76	0.91	0.93
	50	0.93	0.95	0.71	0.92	0.94
	100	0.93	0.95	0.67	0.94	0.94
	200	0.94	0.95	0.62	0.95	0.94
Precision	10	0.93	0.96	0.75	0.92	0.93
	20	0.93	0.97	0.70	0.91	0.94
	50	0.94	0.96	0.64	0.93	0.95
	100	0.94	0.96	0.60	0.95	0.95
	200	0.95	0.96	0.57	0.96	0.95
Recall	10	0.92	0.94	0.89	0.96	0.93
	20	0.92	0.93	0.96	0.90	0.93
	50	0.92	0.94	0.99	0.91	0.93
	100	0.93	0.94	1.00	0.92	0.94
	200	0.93	0.95	1.00	0.94	0.94

Deep Learning (MLP) Results - ECB Mode

Scenario	File Size (KB)	Accuracy	Precision	Recall
S3	10	0.91	0.94	0.89
	20	0.89	0.95	0.91
	50	0.93	0.97	0.93
	100	0.94	0.96	0.92
	200	0.95	0.99	0.92
S4	10	0.90	0.90	0.89
	20	0.90	0.91	0.90
	50	0.92	0.93	0.91
	100	0.93	0.94	0.92
	200	0.94	0.95	0.94

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/ias-2026-0007 | Journal eISSN: 1554-1029 | Journal ISSN: 1554-1010

Journal RSS Feed

Language: English

Page range: 119 - 136

Published on: Jul 8, 2026

Published by: Cerebration Science Publishing Co., Limited

In partnership with: Paradigm Publishing Services

Publication frequency: 6 issues per year

Keywords:

Structures Identification,

Machine Learning,

Symmetric Algorithms,

Cipher Block Chaining Mode,

Cryptanalysis

Related subjects:

Computer sciences,

Fundamentals of computer sciences,

Theoretical computer sciences,

IT-security and cryptology

© 2026 Uroosa Kiran, Hammad Tanveer Butt, Zunera Jalil, published by Cerebration Science Publishing Co., Limited
This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License.

Volume 21 (2026): Issue 3 (July 2026)

Towards Automated Identification of Block Cipher Structures Using Machine Learning

Figures & Tables

Figure. 1:

Figure. 2:

Figure 3:

Figure. 4:

Figure. 5:

Figure. 6:

Figure. 7:

Figure. 8:

Figure. 9:

Figure. 10:

Binary Classification Results for S1_

Average Number of Words against File Sizes

Experimental Scenarios

Diverse Dataset for ML Models

Binary Classification Results for S2

Deep Learning (MLP) Results - CBC Mode

Ciphers Used in Dataset Construction

Binary Classification Results for S3

Binary Classification Results for S4

Deep Learning (MLP) Results - ECB Mode

Paradigm

My account