home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!uwm.edu!uwm.edu!usenet
- From: rick@ee.uwm.edu (Rick Miller)
- Newsgroups: sci.crypt
- Subject: Re: Another well-intentioned novice's question
- Date: 4 Jan 1993 11:23:22 GMT
- Organization: Just me.
- Lines: 40
- Message-ID: <1i96naINN6bo@uwm.edu>
- References: <1993Jan04.051300.26089@rat.csc.calpoly.edu>
- NNTP-Posting-Host: 129.89.2.33
-
- rteasdal@polyslo.csc.calpoly.edu (Rusty) writes:
- >
- > A few months ago, in the letters section of the
- >_Communications of the ACM_, there was published a letter which
- >commented that, in the opinion of the author, cryptographic
- >efficiency would be greatly enhanced by compressing a message
- >before encrypting it, the better to remove repetitive patterns
- >from the plaintext.
- >
- > I don't see, though, that standard (i.e. LZW) compression
- >algorithms actually do remove any of the redundancy from the text;
- >they merely encode and substitute for lengthy repetitive patterns
- >with smaller ones. Mathematically, the whole of the message is
- >still there, is it not?
- >
-
- I'm no "crypto wizard", but what I know of cryptology and compression will
- probably suffice. (Correct me if I'm wrong!)...
-
- In general, compressing your data before encrypting it will do *two* things
- to increase security. First, it makes the resulting message SMALLER.
- Second, it makes the cleartext more RANDOM. Both of these modifications
- will make your communications more difficult to "crack".
-
- SMALLER: Less ciphertext means less hard data for the cracker to work with.
- Much of cryptanalysis is *statistical*, and the accuracy of statistical
- data increases with the amount of data to be analyzed. Less data means
- more *guessing* for the cryptanalyst.
-
- RANDOM: Patterns in the plaintext can sometimes "show through" some algorithms
- into the ciphertext... The more random the plaintext, the less chance
- there is of this happening.
-
- NOTE... It is *STRONGLY* advised that the compression method used
- *NOT* add a standardized 'header' to its compressed output
- (like PKZIP does), as this will make the cryptanalysts' job
- *much* easier.
-
- Rick Miller <rick@ee.uwm.edu> | <rick@discus.mil.wi.us> Ricxjo Muelisto
- Occupation: Husband, Father, WEPCo. WAN Mgr., Discus Sys0p, and Linux fan
-