A new Idea from Gwren of Ants.

[tsafa1]

The following was posted by Gwren in the Mute forum. I will copy and paste here. As you read this please keep in mind that English is not Gwrens primary language. Also please be forwarned that this is not light reading. At times this my get very detailed:

The attached mail shows some ideas I was working on in these days... I'm
testing right now a particular client of Ants that uses the same
protocol of the beta0.7.7 but that can perform a random walk over the
net in order to get nearer and nearer to the sources of its downloads.

Obiously no information exchange happends between nodes, cuz this would
harm privacy. The new client just keep track of which nodes are not
actually useful for download or upload purpose. In practice if you have
some neighbours that only perform routing, that is you are not uploading
your stuff or downloading something through them, those nodes will be
disconnected. This doesn't happend for all your neighbours, cause it
would let some evil node be sure that you are the actual uploader of
something, so only uder certain conditions the disconnection will be
performed, that is only if more than half of your total nodes are not
active, one (randomly) will be disconnected so that the random walk can
take place.

----------------------------------------------------------------
Attached mail from [email protected]
-----------------------------------------------------------------

An idea on how to optimise ANts routing to increase network capacity
without changing ANts protocol but only improving node discovery by
using pheromones.

What do you think?
--------------------------------------------------------------------------------

Increasing Ants Network Capaicty by Reducing Number of Hops Using Pheromones

The connection speed (bandwidth) of a node should not be used to
determine if that node should be connected to. Instead only a
historical record of whether that node has provided or forwarded that
user’s content should be used. As the objective of a node is to file
share (upload or download content) NOT to proxy.

Una rete Multi-hop Ad-Hoc anonima e sicura

The idea for the Ants algorithms is taken from the way in which the ants
behave when searching for food.

Ants first move directly towards the food, but if an obstacle is in the
way between them and the food then they change direction, typically in
an accidental way.

While they move they leave a pheromone trails in their wake.

The concentration of pheromone along a particular trail indicates how
often it was used; in time the pheromone concentration decreases on a
trail if it is not used.

The shortest route to food is used more often, as it takes less time, so
has a stronger concentration of pheromones.

Ants prefer trails with high pheromone concentration so naturally
eventually most will use the shortest route.

Una rete Multi-hop Ad-Hoc anonima e sicura covers:

Route Discovery
Route Maintenance
Route Failure Handling

Una rete Multi-hop Ad-Hoc anonima e sicura does not cover:

Ad-Hoc Overlay Network Capacity
Ad-Hoc Overlay Network Topology
Node discovery

Ad-Hoc Overlay Network Capacity

In the Ants P2P Ad-Hoc Network bandwidth is the scarce resource. It
must be conserved or the network capacity is adversely affected.
Bandwidth is consumed for every hop on network path between two
communicating nodes. To increase capacity the number of hops has to be
reduced.

Say a network has two nodes i and j with bandwidth Bi and Bj then the
available bandwidth is the lower of Bi and Bj.

Say a network has three nodes i, j and k each with same bandwidth B then
the available bandwidth for a proxy connection (two hops) per node: 1/2 x B.

If the minimum optimal path* length is L (number of hops) and number of
nodes is N:

Then for large N see (http://polymer.bu.edu/hes/articles/bbschs04.pdf):

In a random network L scales as N^(1/3)
In a “small world” network L scales as log N
*some paths are direct not proxy

So the capacity for:

A random network scales B/N^(1/3)
A “small world” network scales B/(log N)
Looking at example below it can be seen that network topography does not
significantly affect capacity scaling except when the number of nodes is
large (greater than a 100):

Hops (path length L)
Capacity (B)

No. of Nodes(N)
Random
Small World
Random
Small World

5
1.7
0.7
0.585
1.431

10
2.2
1.0
0.464
1.000

20
2.7
1.3
0.368
0.769

30
3.1
1.5
0.322
0.677

40
3.4
1.6
0.292
0.624

50
3.7
1.7
0.271
0.589

100
4.6
2.0
0.215
0.500

200
5.8
2.3
0.171
0.435

500
7.9
2.7
0.126
0.371

1,000
10.0
3.0
0.100
0.333

1,000,000
100.0
6.0
0.010
0.167

Ad-Hoc Overlay Network Topology

From the above it can be seen that clustering nodes, for example by
connection speed, without taking into account likelihood of file sharing
only marginally increases network capacity for N less than 100.


To increase network capacity the number of hops between file sharing
nodes must be significantly reduced.

This can be done by ensuring that nodes that have a high probability of
sharing content are close together in the network.

ANts needs to use pheromones to discover nodes as well as to discover
routes.

ANts needs to separately record pheromone (content proximity pheromone)
on node IPs when the user’s content is successfully sent or received as
well as when another node’s content is successfully proxied (route
pheromone).

This node IP “content proximity pheromone” needs to be saved between
ANts’ file sharing sessions.

Node IPs that have successfully proxied the user’s content to or from
the user’s node will have a much higher concentration of “content
proximity pheromone”.

Like route pheromone, “content proximity pheromone” should have a TTL
(Time To Live) that decreases as ANts runs.


Users should be more likely to connect to node IPs that have a
historically high concentration of “content proximity pheromone” as
recorded in that user’s records.


The resulting network topology would cluster nodes that are likely to
file share together so dramatically reducing number of hops and so
increasing network capacity.


The average number of hops will never be one (except in very small
network) and so some content will always be proxied.


Also Ants’ ID still will not be able to be linked to IP.


Therefore network will still be anonymous (retain deniability).

Node discovery


The connection speed (bandwidth) of a node should not be used to
determine if that node should be connected to. Instead only historical
record of “content proximity pheromone” should be used as the objective
of the node is to file share (upload or download content).


So slow nodes that historically have proxied the user’s content to and
from their node are better to connect to, than fast nodes that have only
proxied content for other nodes.


The probability of connecting to a node should be weighted by the
“content proximity pheromone”.


The majority of a node’s neighbours will then be nodes that are near
desired content or users wanting that user’s content. However, some
will be random nodes as probability is used. The resulting network
topology will be similar to a “small world” or “power law” topology
(content clusters with a small number of random non content related links).


The “content proximity pheromone” with TTL should also be used to
determine if a neighbour should be disconnected.

Neighbours which do not proxy content to or from a user due to lack of
bandwidth should be disconnected when the “content proximity pheromone”
falls to zero.

NMAX can still be used in node discovery.

Faster connections can have a higher number of neighbours as they have
the capacity to proxy more content to or from neighbours.


Slower connections have less capacity so would only be able to support
fewer nodes.

The ability of a node to upload and download would therefore increase as
the speed of the node increase.


Note, as a node cannot determine if data is own or network traffic prior
to the receipt of the data, the users selfish intention to connect to
other nodes that produce most own traffic benefits the network as a whole.

[tsafa1]

I will be the first to admit that while i understand what he is trying to do (reduce numbr of hops), i have no understanding of how he proposes to do it. If someone does plese tell us.

[Hornet]

I've read it (the above) and I think I understand it.

Gwren has found a way to make Ants download alot faster and to scale to a bigger network without comprimising privacy or speed.

I sure hope this idea works cos if it does it will make ANts a viable alternative to current P2P clients.

Anyway, great to see progress been made at the bleeding edge of P2P development especially after all of today's MPAA raids.

ANts is such a simple idea but it is effective and still has plenty of scope for development.

Can't wait for next giant step :angel

HORNET

[tsafa1]

I just don't understand how these random walkers work to shorten the number of hops.

[crackerjacker]

hmm seems interesting .
btw there is a way to obtain the ip address of the actual host sending the data, in fact there are several ways to do it and several tools to get this information.

hmm
well i might add more to this thread as i can.

[hawkburn]

The NSA and CIA probably have ways of indentifying an individual masked through proxies or "hopping techniques" that the regular person would never even dream of. If the heads of the MPAA and RIAA sleep with enough government people, they might just find the ways as well.

[crackerjacker]

The NSA and CIA probably have ways of indentifying an individual masked through proxies or "hopping techniques" that the regular person would never even dream of. If the heads of the MPAA and RIAA sleep with enough government people, they might just find the ways as well.

Thats a good point, dam there are other ways to man.
hmm there are several software tools that can pave the way to gather information from any p2p or whatever.
hmm
on a side note which has nothing to do with this topic, but is of importance, learning and figuring out the way something operates is the key to finding out the truth behind the core.

i believe that if someone wants to obtain information on many aspects of it, they will be able to. I believe if u search and research and use your intelligence you can easily figure just about anything.

laterz

[tsafa1]

carjacker, your a real idiot. It's not a secret who is on the ants network. They can see the real IP address of everyone via the non-anonymous chatroom. Then, they can could use helicopters to simotaniously raid all 30 people in the chat room at the same time. And they would too, if it was a matter of national security , BUT IT IS NOT. We are not trading nukes, we are trading mp3's and mpg's. This is a cival matter handled by cival courts through cival proceedings, hence the cival lawsuits by the RIAA. The FBI has said that they are not going to spend too many resources tracking down people stealing mp3's. The bit-torrent raids you are seeing is happening because a small number of location providing the means for a large number of people to violate copyright law. These location are relatively few and easy to attack with few resources. They would not be doing it if it was not so easy for them to do in a short time and at little expense. They don't raid the home of every kazaa user because they are too many and would yeild too few results at a very large cost. They ignore kazaa users , thus the RIAA has resorted to a small number of civil lawsuits.

So could the FBI, NSA, CIA, KGB use super computer to triangulate the probable location of Ants users and somehow tie their virtual IP to their real IP and then execute them and dispose of their bodies, I'm sure. Are they gona divert the satalites covering Iraq to do it, I don't think so. So we are back to situation where this ends up as a civil law case, where Denyability is enough.

Now go back to school and get your GED.

[tsafa1]

p.s. if you want to proove your not an idiot tell us exactly how you would tie the ip address to the virtual address, using your "software tools", that non of us know about.

[crackerjacker]

carjacker, your a real idiot. It's not a secret who is on the ants network. They can see the real IP address of everyone via the non-anonymous chatroom. Then, they can could use helicopters to simotaniously raid all 30 people in the chat room at the same time. And they would too, if it was a matter of national security , BUT IT IS NOT. We are not trading nukes, we are trading mp3's and mpg's. This is a cival matter handled by cival courts through cival proceedings, hence the cival lawsuits by the RIAA. The FBI has said that they are not going to spend too many resources tracking down people stealing mp3's. The bit-torrent raids you are seeing is happening because a small number of location providing the means for a large number of people to violate copyright law. These location are relatively few and easy to attack with few resources. They would not be doing it if it was not so easy for them to do in a short time and at little expense. They don't raid the home of every kazaa user because they are too many and would yeild too few results at a very large cost. They ignore kazaa users , thus the RIAA has resorted to a small number of civil lawsuits.

So could the FBI, NSA, CIA, KGB use super computer to triangulate the probable location of Ants users and somehow tie their virtual IP to their real IP and then execute them and dispose of their bodies, I'm sure. Are they gona divert the satalites covering Iraq to do it, I don't think so. So we are back to situation where this ends up as a civil law case, where Denyability is enough.

Now go back to school and get your GED.

Look your mother is the idiot, and so are you.
Fuck your chat room, and fuck you.

YOu go get your ged asshole.
Now back to the real issue at hand, dont worry about what i know.
just because you dont know what i know and swear to god that ants is anoymous when it isnt doesnt bother me.

matter of fact i didnt even come off calling you names i stated my opinion in a civil manner.
i hope your mom dies and i hope u die.
later bitch

[The Hunter]

tsafa1 also no more name calling. Consider this a warning.