home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.crypt
- Path: sparky!uunet!newsgate.watson.ibm.com!yktnews!admin!rs19328!lwloen
- From: lwloen@rchland.vnet.ibm.com (Larry Loen)
- Subject: Re: Attack Methods
- Sender: news@rchland.ibm.com
- Message-ID: <1992Nov20.162319.24428@rchland.ibm.com>
- Date: Fri, 20 Nov 1992 16:23:19 GMT
- Reply-To: lwloen@vnet.ibm.com
- Disclaimer: This posting represents the poster's views, not necessarily those of IBM
- References: <1992Nov18.190513.10997@cis.uab.edu> <1992Nov18.203413.11509@rchland.ibm.com> <1992Nov18.224350.11512@cis.uab.edu> <1992Nov19.215029.22401@bernina.ethz.ch>
- Nntp-Posting-Host: rs19328.rchland.ibm.com
- Organization: IBM Rochester
- Lines: 36
-
- In article <1992Nov19.215029.22401@bernina.ethz.ch>, caronni@nessie.cs.id.ethz.ch (Germano Caronni) writes:
- |> You are discussing the act of padding an encrypted text so, that there
- |> are no peaks in any statistic whatsoever. How about encrypt the whole text,
- |> analyze character-frequencies, and sequences of characters, and then pad
- |> the text in predefined places with characters that flatten the peaks or create
- |> wrong peaks in some sorts of statsistic you would like to apply ? As long as
- |> the opponent does not *know* you have added these pads, perhaps this will
- |> work ?
- |> Other ideas ?
- |>
- |> gec
- |> --
- |> Instruments register only through things they're designed to register.
- |> Space still contains infinite unknowns.
- |>
- |> Germano Caronni caronni@nessie.cs.id.ethz.ch
-
- But, in the real world, the opponent will quickly find out.
-
- Consider the press release. It is sent, in encrypted and padded form
- to 100 branch offices of Fortune 500 firm X. The opponent intercepts
- some of these.
-
- A day later, the plain text is published verbatim, freely available.
- The opponent counts the bytes and goes "aha".
-
- This sort of thing is not uncommon and would reveal the padding pretty
- readily; deciding which text was the press release is a separate problem,
- but 100 copies of the same thing is a pretty good clue. . .and this
- exact kind of attack _has_ succeeded many times. Government crypto
- groups spend a lot of time on it. They can often correlate very
- accurately even when the can't solve the cipher.
-
- --
- Larry W. Loen | My Opinions are decidedly my own, so please
- | do not attribute them to my employer
-