Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#15 Threshold pool feature

closed-rejected
nobody
None
5
2002-10-27
2002-10-03
Trek Star
No

This patch add the threshold feature (and configuration
option) as described in "From a Trickle to a Flood:
Active Attacks on Several Mix Types"
<http://freehaven.net/doc/batching-taxonomy/taxonomy.ps>
to improve the minimum anonymity.

Now RATE operates on all messages minus the POOLSIZE
and not on all messages present in the pool. Also a
sanity check in RATE option is maked and the default
value is changed to 90 (from 95). It should be smaller
(like 70-40).

NOTE: I don't have all the current CVS tree, then i
can't test it.

Discussion

  • Trek Star
    Trek Star
    2002-10-03

    patch for the current CVS tree (20021002)

     
  • Trek Star
    Trek Star
    2002-10-04

    second patch for the CVS tree (20021002)

     
  • Trek Star
    Trek Star
    2002-10-04

    Logged In: YES
    user_id=365839

    I'm not confident with my previus patch, because i don't
    understand all of the code; here attached a new smaller
    version that only add threshold and don't make any sanity
    check; THRESHOLD is not exactly the threshold to send pool,
    but is threshold-1 (because we make a <= check and not a <
    in pool:432).

    Now what I don't understand in pool.c:445:

    max = size * RATE / 100; // here RATE is calculated on
    all messages
    if (max < 0) // this check is useful ONLY if RATE is
    negative.. true?
    max = 1;

    well if we make a sanity check on RATE, we can remove the
    check (max < 0); then I must understand if RATE should be
    applied on all messages (size) or only to size-POOLSIZE
    messages.

    Finally some sanity check could be done with a special
    read_conf_i() that takes minimal and maximal values too.

     
    • status: open --> closed-rejected
     
  • Logged In: YES
    user_id=1525

    Roger did not really convnce me that the additional
    threshold parameter increases anonymity as an operator could
    always just increase the poolsize anyway.

    Therefore I do not think changing and complicating our
    mixing strategy will have any bonus and reject the patch.