[freehaven-cvs] i could fiddle with this forever

[arma@seul.org (Roger Dingledine)]

Update of /home/freehaven/cvsroot/doc/sync-batching
In directory moria.mit.edu:/home2/arma/work/freehaven/doc/sync-batching

Modified Files:
	sync-batching.pdf sync-batching.tex 
Log Message:
i could fiddle with this forever
so i'm going to submit it instead


Index: sync-batching.pdf
===================================================================
RCS file: /home/freehaven/cvsroot/doc/sync-batching/sync-batching.pdf,v
retrieving revision 1.17
retrieving revision 1.18
diff -u -d -r1.17 -r1.18
Binary files /tmp/cvs5ortnr and /tmp/cvse1c07O differ

Index: sync-batching.tex
===================================================================
RCS file: /home/freehaven/cvsroot/doc/sync-batching/sync-batching.tex,v
retrieving revision 1.50
retrieving revision 1.51
diff -u -d -r1.50 -r1.51
--- sync-batching.tex	27 Jan 2004 14:20:30 -0000	1.50
+++ sync-batching.tex	27 Jan 2004 15:43:19 -0000	1.51
@@ -581,9 +581,9 @@
 For $m=128$ messages and $u=4$ nodes (i.e.~cascade or stratified network),
 the chance of any node getting less than 16 messages (compared to
 the 32 we expect each to get) is $6 \cdot 10^{-4}$---meaning with very high
-probability the average entropy approximates the behavior we will see
-in reality. On the other hand, for $u=16$ nodes (free-route), 48\% of
-the time some node will get less than half the expected number. It is
+probability the average entropy represents the behavior we will see
+in reality. However, for $u=16$ nodes (free-route), 48\% of
+the time some node will get less than half the expected number; and it is
 not until a batch size of 480 that this metric reaches 1\%.
 
 This result makes sense: each link on a free-route network has a smaller
@@ -592,11 +592,11 @@
 expect batches to be in reality? The Mixmaster network receives more than
 1000 messages an hour, which seems plenty sufficient.
 Second, how bad is it when a link varies by half the expected volume? If
-we change our metric to require at least 2 messages on each link for
-$m=128$, then we find that only 1\% of the cases fall outside this value.
+we change our metric to require at least 2 messages on each link, then for
+$m=128$ we find that only 1\% of the cases fall outside this value.
 Another significant question is how the number of layers affects the
 results: the more layers, the greater the chance that some of them are
-well balanced. But, the exact relation and its effect on entropy are
+well balanced. The exact relation and its effect on entropy are
 open questions.
 
 Danezis also considers this issue of variance from average entropy for his

***********************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe freehaven-cvs       in the body. http://freehaven.net/