Encryption Algorithm Identification through ML with Character Frequency Approach
Guruprasad Dhawade, Aditya Kale, Aman Gupta, Iliyas Sayyad, Kishor S. Wagh · 2025
In an era where data security is paramount, identifying encryption algorithms used in intercepted data becomes critical for cybersecurity analysts. This research presents a novel machine learning-based framework leveraging character frequency analysis to identify the underlying encryption technique. As data security increasingly relies on cryptographic algorithms, it becomes imperative to develop effective methods for identifying and analyzing these algorithms to ensure their robust implementation. This research utilizes Machine Learning approaches, such as Support Vector Machine (SVM), Naïve Bayes, Decision Tree, and Random Forest, to categorize encryption algorithms by analyzing ciphertext features. The results demonstrate varying performance across models, with SVM achieving an accuracy of $\mathbf{8 3 \%}$, Decision Tree at $81 \%$, Random Forest excelling with $85 \%$, and Naive Bayes providing a reasonable but comparatively lower accuracy of $79 \%$. This research presents an innovative method that utilizes machine learning to automatically identify encryption algorithms. The proposed methodology involves analyzing ciphertext samples generated by commonly used encryption schemes, including AES, ChaCha20, and Blowfish. Features such as character frequency distribution in encrypted text are extracted and utilized to train machine learning models, enhancing their ability to distinguish between different cryptographic methods effectively.