The biases with RC4 derived ciphers have to deal with the unlikelihood that an equivalent value (usually zero) will occur near another in the first few bytes of every 256 byte block.
Each byte is equally probable of occurring, though. By randomly permuting the bytes, and scrambling the permutation array after each block, it removes the bias. The original keystream should be sufficiently random to scramble it. I call it a self-scrambling generator and the core concept could be paired with any stream cipher. Might reduce the strength of some related key attacks. I'm interested if a person can show a distinguishing attack against this. On Wed, Aug 20, 2014 at 3:40 AM, Jeffrey Walton <[email protected]> wrote: > On Wed, Aug 20, 2014 at 4:39 AM, Ryan Carboni <[email protected]> wrote: > > Feed RC4 through a transposition cipher... essentially a single round > > 2048-bit block cipher. > > > > Table 1: 256 permuted bytes, serves as the PRGA > > Table 2: 256 permuted bytes, serves as the transposition cipher > > Table 3: 256 empty values, serves as the output array > > Table 4: 256 empty values, serves as the output array to rescramble the > > transposition cipher > > ... > > > > Just wondering if it's a good change. > Wouldn't you still have the same biases, but in different places? >
_______________________________________________ cryptography mailing list [email protected] http://lists.randombit.net/mailman/listinfo/cryptography
