p2psp team mailing list archive

[Vicente Gonzalez <vicente.gonzalez.ruiz@xxxxxxxxx>]

2015-03-18 3:55 GMT+01:00 Ilshat Shakirov <im.shakirov@xxxxxxxxx>:

> Hello!,
>

Hi!


>
> Yes, trusted peer and attacker have always the same position (attacker -
> 1, TP - 2), but this doesn't affect on the result because of random
> latencies. Splitter always have peer list in random order because of
> latencies.
>

I believe you (that the positions of the peer in the list of peers are
random), but I don't think that the reason is that the latencies are also
random. This should affect slightly the order in which the peer's buffers
are filled, but not to the order in which peers send the chunks. This order
(the peer ordering in the list) is responsible of the "protocolol latency":
some peers are going to fill up their buffers almost in a perfect order and
others with large gaps (almost randomly) and this can determine the number
of chunks that the attacker can send before it is detected by the trusted
peer. Sorry, this is the only reason I can imagine that explains why the
simulations do not show the expected results :-/


> Here's log file of experiment with 50 peers.
> https://www.dropbox.com/s/cy2ljkacbs605c3/log_50peers.txt?dl=0 - 100
> experiments, 50 peers, with buffers (48 Mb)
>
> https://www.dropbox.com/s/rmiet1vby1f46hf/avg_posioned_chunks_a1_t1_p50.txt?dl=0
> - 100 experiments, 50 peers, without buffers (167 Kb)
> When I prepared the results, I simply 'grepped' the strings with regex and
> paste it in google sheet.
>

Well, the output seems to be OK, but I cannot see the list of peers.

Best,
Vi.

-- 
Vicente González Ruiz
Depto de Informática
Escuela Técnica Superior de Ingeniería
Universidad de Almería

Carretera Sacramento S/N
04120, La Cañada de San Urbano
Almería, España

e-mail: vruiz@xxxxxx
http://www.ual.es/~vruiz
tel: +34 950 015711
fax: +34 950 015486