Re: [B.A.T.M.A.N.] B.A.T.M.A.N Digest, Vol 61, Issue 9 (UNCLASSIFIED)

Re: [B.A.T.M.A.N.] B.A.T.M.A.N Digest, Vol 61, Issue 9 (UNCLASSIFIED)

[Karan, Cem F CIV (US)]

[-- Attachment #1: Type: text/plain, Size: 1627 bytes --]

Classification: UNCLASSIFIED
Caveats: NONE

> Marek Lindner wrote on Mon, 9 Jan 2012 05:29:26 +0800:
> 
> > I'm a roboticist.  My current project involves creating robots that
> > are mobile ad-hoc routers for a network.  The robots are supposed to
> > autonomously determine where to move to improve the
> > bandwidth/latency/reliability/etc. of a network.  In order for this to
> > work, each robot needs to have a rough approximation of the topology
> > of the network, so they can measure if their movements are improving
> > or worsening the network.  The topology information needs to include
> > not just what the network is currently like (which, from the
> > documentation, appears to always be a tree), but a complete graph,
> > with measured information about the bandwidth/latency/reliability of
> > the links between pairs of nodes.  Is it possible to get this
> > information from batman-adv?
> 
> a single batman node does not have the information about all links and
> their properties (bandwidth/latency/reliability) in the network. The
> reason is quite
> simple: The larger the mesh the more outdated your information will be.

That is actually acceptable and expected.  I just want the information
between the one-hop neighbors to be relatively timely.  The only requirement
for the further away robots that I have is that if the robot positions and
the world are static, then the topology information should more or less
stabilize over time.  Random fluctuations are OK; they can be averaged away
without any trouble.  Does that make sense?

Thanks,
Cem Karan

Classification: UNCLASSIFIED
Caveats: NONE



[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5614 bytes --]

Re: [B.A.T.M.A.N.] B.A.T.M.A.N Digest, Vol 61, Issue 9 (UNCLASSIFIED)

[Marek Lindner]

On Tuesday, January 10, 2012 06:47:47 Karan, Cem F CIV (US) wrote:
> > a single batman node does not have the information about all links and
> > their properties (bandwidth/latency/reliability) in the network. The
> > reason is quite
> > simple: The larger the mesh the more outdated your information will be.
> 
> That is actually acceptable and expected.  I just want the information
> between the one-hop neighbors to be relatively timely.  The only
> requirement for the further away robots that I have is that if the robot
> positions and the world are static, then the topology information should
> more or less stabilize over time.  Random fluctuations are OK; they can be
> averaged away without any trouble.  Does that make sense?

I don't know much about robots or your scenario in particular but wifi 
properties do change even when all participants are not moving. Interference 
can come from any external source you can imagine.

Regards,
Marek