CryptoDetection: A Cryptography Misuse Detection Method Based on Bi-LSTM

Abstract

As the dramatic increase of both number and types of IoT devices, especially the ones have stringent requirements on complexity, development period and cost, cryptography misuses are becoming increasingly common which will have a tremendous influence on the security and privacy protection. To address the above question, this paper studies the feature of 73 cryptography misuses, proposes a cryptography detection method based on Bi-LSTM which can learns forward and backward simultaneously, and then developes a prototype tool called CryptoDetection for Java source code. Experimental results have shown that the proposed detection method achieves 92% accuracy, 92% precision, 91% recall and 92% f1-score, which respectively outperforms the best one among all the compared state-of-the-art techniques.