Large-scale randomization techniques

Abstract

This paper looks at a collection of especially simple conventional cryptosystems that use a very large blocksize. One variation uses a single xor randomization followed by a single bit permutation. Tight upper and lower bounds are obtained on the number of bits of matching plaintext/ciphertext needed to break the systems. These results follow from two interesting combinatorial theorems. The cryptosystems are not practical because the number of bits above is about the same as the keysize. We can make the systems practical by introducing key-dependent pseudo-random numbers, though we then lose any proofs of the difficulty of cryptanalysis.