Combiner-Type Algorithms (original) (raw)
ACiphers By Ritter Page
Comments on codebook and combiner algorithms.
"Codebook" algorithms are block ciphers. "Combiner" algorithms are stream ciphers.
Contents
- 1998-11-02 John Savard: "for 'Combiner style' algorithms to be classed as the other major category of algorithm indicates this field has been rather a busier one behind closed doors...."
- 1998-11-03 Douglas A. Gwyn: "Of course it has!"
- 1998-11-03 John Savard: "I was thinking of Terry Ritter's usage...."
- 1998-11-03 Terry Ritter: "That looks like the same English usage to me: Multiple values are combined into a single result."
Subject: Combiner-Type Algorithms (was: Alternative to the Embassy...) Date: Mon, 02 Nov 1998 18:07:45 GMT From: jsavard@tenMAPSONeerf.edmonton.ab.ca (John Savard) Message-ID: 363df3e2.5879047@news.prosurfr.com References: 71dtci$rtd$1@server.cntfl.com 3639ef90.860249@news.prosurfr.com Newsgroups: sci.crypt,talk.politics.crypto,comp.security.pgp.discuss Lines: 19
"R H Braddam" rbraddam@aic-fl.com wrote, in part:
There are three 32-bit four-stage pipelined RISC processors running at 100MHZ. Two of those are in the Programmable Cryptographic Processor which has a combined throughput of >1200 MIPS. One is optimized for Codebook type algorithms, the other for Combiner type algorithms.
Paging Terry Ritter?
Except for my recent "Large-Key Brainstorm", and Terry Ritter's work with Dynamic Substitution and Latin Squares, I've heard of block ciphers - which could be called "Codebook style" algorithms, and stream ciphers. But for "Combiner style" algorithms to be classed as the other major category of algorithm indicates this field has been rather a busier one behind closed doors...
John Savard http://members.xoom.com/quadibloc/index.html
Subject: Re: Combiner-Type Algorithms (was: Alternative to the Embassy...) Date: Tue, 03 Nov 1998 00:14:40 GMT From: "Douglas A. Gwyn" DAGwyn@null.net Message-ID: 363E4A8E.37399E0B@null.net References: 363df3e2.5879047@news.prosurfr.com Newsgroups: sci.crypt,talk.politics.crypto,comp.security.pgp.discuss Lines: 9
John Savard wrote:
... But for "Combiner style" algorithms to be classed as the other major category of algorithm indicates this field has been rather a busier one behind closed doors...
Of course it has! "Combiner" is the component that takes bits from the state of a stream cipher system and "combines" the bits, usually these days in some nonlinear way, to compute a key bit.
Subject: Re: Combiner-Type Algorithms (was: Alternative to the Embassy...) Date: Tue, 03 Nov 1998 16:30:40 GMT From: jsavard@tenMAPSONeerf.edmonton.ab.ca (John Savard) Message-ID: 363f2efa.273062@news.prosurfr.com References: 363E4A8E.37399E0B@null.net Newsgroups: sci.crypt Lines: 24
"Douglas A. Gwyn" DAGwyn@null.net wrote, in part:
John Savard wrote:
... But for "Combiner style" algorithms to be classed as the other major category of algorithm indicates this field has been rather a busier one behind closed doors...
Of course it has!
I know cryptography in general is busier behind closed doors than otherwise; that's only reasonable, since the major users are those who that serves. I meant that the specific field of that type of algorithm was busier in relative terms - compared to the concentration on block ciphers in public work.
"Combiner" is the component that takes bits from the state of a stream cipher system and "combines" the bits, usually these days in some nonlinear way, to compute a key bit.
Ah. I was thinking of Terry Ritter's usage, where the combiner is what applies the key bits to the plaintext to produce ciphertext.
John Savard http://members.xoom.com/quadibloc/index.html
Subject: Re: Combiner-Type Algorithms (was: Alternative to the Embassy...) Date: Tue, 03 Nov 1998 18:50:40 GMT From: ritter@io.com (Terry Ritter) Message-ID: 363f5057.10172037@news.io.com References: 363f2efa.273062@news.prosurfr.com Newsgroups: sci.crypt Lines: 53
On Tue, 03 Nov 1998 16:30:40 GMT, in 363f2efa.273062@news.prosurfr.com, in sci.crypt jsavard@tenMAPSONeerf.edmonton.ab.ca (John Savard) wrote:
"Douglas A. Gwyn" DAGwyn@null.net wrote, in part: [...]
"Combiner" is the component that takes bits from the state of a stream cipher system and "combines" the bits, usually these days in some nonlinear way, to compute a key bit.
Ah. I was thinking of Terry Ritter's usage, where the combiner is what applies the key bits to the plaintext to produce ciphertext.
That looks like the same English usage to me: Multiple values are combined into a single result.
Combiners can be nonlinear and irreversible, and so can be used to combine RNG sequences to produce a stronger sequence.
Combiners are commonly exclusive-OR, which is linear and reversible; this supports data with running key in a stream cipher, but of course the combining has no strength at all under known-plaintext.
Another is Dynamic Substitution, which is nonlinear yet reversible, and so can be used for stream cipher data / running-key combining, and yet also have strength under known-plaintext.
I know "combiner" from the patent literature, and the earliest references largely describe linear and reversible combining (e.g., exclusive-OR):
We see a vacuum tube exclusive-OR in Smith 2,496,317 (1950 Feb 7) titled "Combining Circuit."
We see another vacuum tube exclusive-OR in Kohler (1951 Sep 11) titled "Combining Circuits."
We also see a mechanical exclusive-OR (for teletype encryption) in Arko 3,159,712 (1961 Aug 30) titled "Mechanical Signal Combiner."
In body text I happened to come upon Blasbalg 3,715,508 (1973 Feb 6) titled "Switching Circuits Employing Orthogonal and Quasi-Orthogonal Pseudo-Random Code Sequences." In col 3, line 28, we see: "Combiner 7 may be of any well known type such as a multiplier for sampled analog data or a modulo-two adder for digital data."
Terry Ritter ritter@io.com http://www.io.com/~ritter/ Crypto Glossary http://www.io.com/~ritter/GLOSSARY.HTM
Terry Ritter, hiscurrent address, and histop page.
Last updated: 1999-01-19