Encryption scheme classification: a deep learning approach

Abstract

Encryption has an important role in protecting cyber assets. However use of weak encryption algorithms could render this intent useless as it could be exploited to gain unauthorized access to these important assets. This vulnerability may be exploited intentionally. Hence this vulnerability has been formally recognized with its own Common Vulnerabilities and Exposures (CVE) label by cyber security community as the vulnerability to protect. When exploited, detecting this vulnerability from encrypted data is very difficult task to undertake. This research explores the use of recent advancement in machine learning algorithms specifically deep learning algorithms to classify encryption schemes based on entropy measurements of encrypted data with no feature engineering. Past research work using various machine learning algorithms have failed to achieve good accuracy results in classification. The research entails applying encryption algorithms Data Encryption Standard (DES) and Advanced Encryption Standard (AES) with block cipher modes namely Electronic Codebook (ECB) and Cipher Block Chaining (CBC) over the image dataset from CIFAR10. Two ImageNet winning Convolutional Neural Network deep learning models namely AlexNet and GoogleNet are used to perform the classification. Transfer learning and layer modification were applied to evaluate the classification effectiveness. This research concludes that deep learning algorithms can be used to perform such classification where other algorithms have failed.