Repetition Pattern Attack on Multi-word-containing SecureString 2.0 Objects

Cloud computing appeals to private individuals and particularly enterprises at a progressive rate, but a noticeable percentage of them refuse it due to mistrust or missing commitment to security. The cryptosystem SecureStringi¾ź2.0 was designed to outweigh these deficits through the support of blind computations on character strings. Repetition pattern attacks count among the most hazardous enemies of SecureStringi¾ź2.0 objects because reoccurring ciphergrams within them may reveal linguistic identifying features of the correspondent plaintext. This paper analyzes and compares the success probability of repetition pattern attacks on the following three sorts of SecureStringi¾ź2.0 objects: single-word-containing ones, multi-word-containing ones with a known number of words plus unknown delimiter positions, and multi-word-containing ones with an unknown number of words plus unknown boundary locations. The latter type is expected to provide the highest privacy.