Evaluation of Encrypted Data Identification Methods Based on Randomness Test
Bo Zhao, Qinrang Liu, Xiaomin Liu · 2011
Randomness tests have been employed in encrypted data identification since encrypted data have high randomness. However, the evaluation of various randomness tests is absent. The NIST statistical test suite is performed on encrypted and compressed data, which are derived from seven different types of files. And an index, differentiability, is introduced to indicate the ability of a test to identify the encrypted data. Results show the effect of differentiability, and suggest that cumulative sums test have the best differentiability. Furthermore, some compressed data are similar to encrypted data and the identification of them is a major point of future work.