🚀 go-pugleaf

Howdy,
	I have successfully incorporated Dawson and Tavares Expanded
design [1] and Adams S-boxes [2] to build a 64 bit key DES which uses
the 16 sub-keys to pick out 16 blocks of 16 4x4 S-boxes.  I can't
believe it works!  This seems so complex I'm not sure I can describe
it.

	I typed in the 57 S-boxes that Carlisle sent me.  I used the
Dawson tables to expand my working version of DES into a 64 bit key.
For each subkey, I use the first 7 bits to determine where to "split"
the remaining 57.  I then alternatly scan up and down from this point
(assuming the hamming weight was > 16) and use the index for each bit
that is set as an index into Adams S-boxes.  The row of 16 bytes (one
4x4 S-box) is copied into next slot in the master S-table until 16
rows have been filled.

	For each round, I use the (round index * 128) as the
starting point in the master S-table for each F module's 8
Substitution blocks. The round index is reversed between encrypting
and decrypting and it actually works both forwards and backwards
(encrypting and decrypting).  

	I also put in a set of random numbers to build an 8x4 S-box
from the 57 4x4 list in case the Hamming weight of a key was less than
16.

	Thanks to everyone who responded to my initial requests for
information.  I've had a blast getting this to work.

[1]  "An Expanded set of design criteria for substitution boxes and
their use in strengthening DES-like cryptosystems", Dawson, M.H.;
Tavares, S.E.; IEEE Pacific Rim Conf. on Communications, Computers
and Signal Processing, Victoria BC May 1991, p191

[2]  private communication from Carlisle Adams

Patience, persistence, truth,         reality:       mgr@anhep2.hep.anl.gov
Dr. mike                                 UUCP:    uunet!pyramid!cdp!mrosing
IMI, P.O. BOX 2242, Darien IL 60559    bitnet:              mrosing@igc.org