p2psp team mailing list archive

[Ilshat Shakirov <im.shakirov@xxxxxxxxx>]

Hello,

I performed a small experimens, results are here:
https://docs.google.com/spreadsheets/d/1pc6yb87xJy8gNkWSWvvCvjAjR6WBFdOzCbPvO-zEooU/edit?usp=sharing


(On first tab of document is raw data from simulator).
Few words about experiment - I performed it 100 times for each network
size. Experiment was performed with 1 trusted peer and 1 attacker.
Simulation value of poisoned chunks simply is average from all experimens.

You can see that simulation value is always greater than P / 2. I think it
should be taken into account. Am I wrong? What do you think?

PS. What parameters should I use in the next experiments? Im about number
of trusted peers, attackers, network size. Thanks in advance.

2015-03-17 4:21 GMT+06:00 Vicente Gonzalez <vicente.gonzalez.ruiz@xxxxxxxxx>
:

>
>
> 2015-03-16 15:28 GMT+01:00 Ilshat Shakirov <im.shakirov@xxxxxxxxx>:
>
>> Hello!,
>>
>> In the simulation, the peer number 1 is an attacker?
>>
>> Yes, the peer number 1 is malicious peer, peer number 2 is trusted peer.
>>
>> Btw, I am little bit confused about average number of poisoned chunks
>> (P/2, T=1, A=1). It will be P/2 if all the peers immediately get to know
>> about malicious peer after the trusted peer receives poisoned chunk. But it
>> can't be, trusted peer must send the poisoned chunk to the splitter and
>> splitter must send message to the rest of team about malicious peer. While
>> trusted peer and splitter messaging, malicious peer will send poisoned
>> chunks to the team. I think average number of poisoned chunks in the team
>> will be P/2 + C, where C depends on network bandwidth.
>>
>
> Well, obviously the latency of the system (including bandwidth issues) are
> going to increase the number or poisoned chunks, but I still think that in
> average the number of poisoned chunks should be #T/2, where #T is the
> number of peers in the team T. However, the simulation will tell us if I am
> right or no :-)
>
>
>>
>> I will implement the consecutive sending chunks, and will perform many
>> experiments to evaluate this value.
>>
>
> Thanks a lot!!
>
> Best,
> Vicente.
>
>
>>
>>
>> 2015-03-16 2:47 GMT+05:00 Vicente Gonzalez <
>> vicente.gonzalez.ruiz@xxxxxxxxx>:
>>
>>>
>>>
>>> 2015-03-13 17:56 GMT+01:00 Ilshat Shakirov <im.shakirov@xxxxxxxxx>:
>>>
>>>> Hello!,
>>>>
>>>
>>> Hi IIshat!
>>>
>>>
>>>>
>>>> I just pushed the new version of PeerSIm impl of p2psp-protocol
>>>> (lp:~im-shakirov/p2psp/sim).
>>>> I have implemented the malicious peers and the CIS of rules (model 1,
>>>> STrPe). New log file at the same place -
>>>> https://www.dropbox.com/s/dh1cewia6ahyoiy/log.txt?dl=0
>>>>
>>>>
>>> I see. Good work. Only one question. In the simulation, the peer number
>>> 1 is an attacker?
>>>
>>>
>>>> The first results is that there is some problems to acheive average
>>>> number of poision chunks at P / 2 (P is number of peers, as described at
>>>> http://p2psp.org/en/p2psp-protocol?cap=indexsu11.xht#x20-160004.11)
>>>> There is always P - 1 poisoned chunks in simulation. But to be honest,
>>>> now peers send their chunks to the rest of team simultaneously (as
>>>> described in fact 1 -
>>>> http://p2psp.org/en/p2psp-protocol?cap=indexse2.xht#x4-30002).
>>>>
>>>
>>> Well, the granularity that you are using in the simulation (100 steps of
>>> time) can be insufficient to simulate the behaviour we predict in this
>>> document (that in average, the half of the peers are going to be poisoned).
>>>
>>>
>>>>
>>>> What do you think? Should I to implement the process of sending chunks
>>>> (from peers) as consecutive actions? In other words the peer sends
>>>> chunks one by one. If so, there is can be acheived the P / 2 number of
>>>> poisoned chunks, I think.
>>>>
>>>>
>>> I think so :-)
>>>
>>> Regards,
>>> 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
>>>
>>
>>
>
>
> --
> 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
>