Efficient and provably secure ciphers for storage device block level encryption

Block ciphers generally have fixed and relatively small input length. Thus they are often used in some mode of operations (e.g., ECB, CBC, CFB, and CTR) that enables the encryption of longer messages. Unfortunately, all these modes of operation reveal some information on their inputs or on relationships between different inputs. As an example, in the CBC mode, encrypting two messages with an identical prefix will result in identical initial blocks in the ciphertexts. Due to the well-known birthday attack and the small input length, the CBC mode becomes less secure as the number of data blocks to be encrypted increases. This leads to a challenging task, namely to design schemes for storage device block or sector level data encryption that are efficient and do not have the disadvantages mentioned above. In this paper, we propose an efficient cipher whose data/cipher blocks can be specified flexibly to match the length of a block unit for current and foreseeable future storage devices. We show that our encryption scheme is provably secure under the assumption that the underlying one-way hash function is a random function.