Expunge revision histories in files.

[u/mdw/catacomb] / papers / rand.tex

diff --git a/papers/rand.tex b/papers/rand.tex
index eda970e..d4ae0c1 100644 (file)
--- a/papers/rand.tex
+++ b/papers/rand.tex
@@ -1,6 +1,6 @@
 %%% -*-latex-*-
 %%%
-%%% $Id: rand.tex,v 1.2 1999/10/12 21:00:34 mdw Exp $
+%%% $Id: rand.tex,v 1.4 2004/04/08 01:36:15 mdw Exp $
 %%%
 %%% Description of Catacomb's random number generator
 %%%
@@ -26,16 +26,6 @@
 %%% Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
 %%% MA 02111-1307, USA.
 
-%%%----- Revision history ---------------------------------------------------
-%%%
-%%% $Log: rand.tex,v $
-%%% Revision 1.2  1999/10/12 21:00:34  mdw
-%%% Updated.  Almost finished, in fact. ;-)
-%%%
-%%% Revision 1.1  1999/09/03 08:41:13  mdw
-%%% Initial import.
-%%%
-
 %%%----- Header -------------------------------------------------------------
 
 \documentclass[a4paper, article, 10pt, notitlepage, numbering]{strayman}
@@ -200,13 +190,24 @@ The Catacomb implementation of the generator uses the following parameters:
 \item The hash function used in the one-way transformation is RIPEMD-160
   \cite{rmd160}; the block cipher is Blowfish, using a 160-bit key.
 \item The input pool size $N_I$ is 128 bytes.  The output buffer size $N_O$
-  is also 128 bytes.  The size $N_S$ of the secret part of the output buffer
+  is 512 bytes.  The size $N_S$ of the secret part of the output buffer
   is 160 bits (20 bytes).
 \item The polynomial $P(x)$ used for mixing in new input is
   $1 + x + x^2 + x^7 + x^{128}$.
 \end{itemize}
 The hash and block cipher are well-known and respected cryptographic
-primitives.  
+primitives.
+
+The input pool is rater larger than it strictly needs to be to contain
+`enough' entropy to bring the generator up to the strength of its
+cryptographic primitives.  The pool is large to reduce the effect of
+asymptotic behaviour in the amount of entropy in the pool.
+
+The output buffer is large simply to improve performance: Blowfish has a
+heavy key schedule, so it pays to perform fewer rekeyings per byte of data.
+The precise size of 512 bytes was chosen empirically as being about where the
+performance improvement stops being linear with the buffer size on my
+machine.
 
 \begin{thebibliography}{99}