home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!destroyer!cs.ubc.ca!unixg.ubc.ca!unixg.ubc.ca!unruh
- From: unruh@unixg.ubc.ca (Bill Unruh)
- Newsgroups: sci.crypt
- Subject: Re: DES Encryption/ Encrypting more than once.
- Date: 15 Oct 92 17:23:49 GMT
- Organization: The University of British Columbia
- Lines: 24
- Message-ID: <unruh.719169829@unixg.ubc.ca>
- References: <wa6JsB7w165w@works.uucp> <1992Oct13.174505.24230@b11.b11.ingr.com> <1992Oct15.125830.25539@bnr.ca>
- NNTP-Posting-Host: unixg.ubc.ca
-
-
-
- keith@bnr.ca (Keith W. Campbell) writes:
-
-
- >It is only a problem if the cycle length is very short. A cycle length of
- >one means that the plaintext and ciphertext are the same. If the cycle
- >length is two, an attacker who can arrange to have a chosen text encrypted
- >may discover the plaintext by having the ciphertext re-encrypted.
-
- >For DES these short cycles are very rare and such keys are easy to avoid.
- >DES and other good ciphers which may be considered to be random mappings
- >have average cycle lengths of sqrt(pi*n/8) steps where n is the number of
- >possible plaintexts.
-
- >For DES, n=2^64 so the average cycle length is about 0.627*2^32.
- >So in practice, the cycles of an
- >encryption algorithm are not problematic.
-
- So a chosen plaintext attack of "only " 2^32 should crack DES, rather
- than the 2^47 or something of the differential analysis attack? Sounds
- like a significant advance in the breaking of DES since the former has
- been widely reported as breaking DES:)?
-
-