On Tue, Sep 11, 2007 at 11:53:39AM +0200, <ZioPRoTo (Saverio Proto)>:
~> 1)
~> Is there any document where the packet structure of the TP messages is
defined?
~> I believe it would help the comprehension of the section three "tracer
~> packet", in the Quantum Shortest Path Netsukuku Paper.
No, there isn't. You should read the source code, because it varies from
implementation to
implementation. The paper is abstract, it shouldn't mingle with implementation
details.
However, in python, you can think of a list [] of node IDs, f.e.
[ 51, 2, 11, 29, 120 ]
~> First of all it is not clear if the same packet structure is shared by
~> TP, ATP, and CTP floods.
Yes, they are all TP, but each handled in a different way.
~> 2)
~> Also it is not very clear how the CTP (continuos tracer packet) does
~> not collapse. If a CTP is a TP flood which will never ends seems like
~> having packets looping in the network forever.
~> Is the expiring of the packets completely regualted only by the
~> Interesting Information Rule?
Note that the CTP isn't completely used. It's just described. The ETPs do all
the work.
In the previous version of qspn.pdf, when the ETP didn't exist, we used
CTP+Interesting Information Rule+MaxRoute, and we showed that this was good to
explore the network (see this old version:
http://hinezumi.org/viewcvs/netsukuku/doc/main_doc/qspn.pdf?revision=1.14 )
~> 3)
~> Moreover what is the maximum size of a TP packet ?? (this answer is
~> straightforward to the definition in the paper of the TP packet)
A gnode has a maximum of N nodes.
The worst case is when all the gnode are in a segment:
node 1 -- node 2 -- ... - node N
so, maxTPsizez = sizeof(node_id)*N
in the current implementation N = 256
node_id == char ==> sizeof(node_id) = 1 ==> maxTPsizez = 256
However, this is just a pure TP. You should consider also ancillary data.
~> 4)
~> The last concern is about the use of RSA to sign the routes containted
~> in the TP, isn't the validation of the signature a too heavy operation
~> in terms of computational burden for a node that needs to scan many TP
~> packets?
The number of ETPs should be small, unless the described theory isn't valid.
The coming simulations will show approximately how large this number is.
~> Better crypto schemes different than RSA exist based on
~> public key cryptography. For instance sensor networks that are
~> oriented on low CPU consumption are adopting elliptic curves public
~> key crypto systems.
I don't know. Can you post some links?
Important Note: for the current version of the whole theory, no security
issues have been taken into consideration, because for now we are aiming to
make it just work. Thus, the "Crypto QSPN" section is more a TODO than
something fully developed.
--
:wq!
"I don't know nothing" The One Who reached the Thinking Matter '.'
[ Alpt --- Freaknet Medialab ]
[ GPG Key ID 441CF0EE ]
[ Key fingerprint = 8B02 26E8 831A 7BB9 81A9 5277 BFF8 037E 441C F0EE ]
_______________________________________________
Netsukuku mailing list
[email protected]
http://lists.dyne.org/mailman/listinfo/netsukuku
