ORBIT-USER: Setting up Topologies

[Sanjit Kaul]

Using an example I will try to explain my understanding of your question and
give some insight into what is possible. The scale of my example is much
smaller than what you seem to suggest. To know what is possible at the scale
you are suggesting we can thrash out the details of your requirement and
evaluate feasibility of mapping. It is difficult to make a blanket statement
when 15 - 20 nodes are involved.

Topology:
--------------

Consider, three nodes A, B, C. Consider a topology where

1) B cannot decode C's transmissions but can decode A's transmissions.
2) C cannot decode either A or B's transmission in case the transmissions
occur simultaneously.
3) Also, A and B are outside each others carrier sense range.

The topology can be mapped on the grid. For 3) to happen we need to modify
the carrier sense threshold of the card. It is an issue we are currently
looking at.

Please let know if the example demonstrates what you are looking for.

Thanks,
Sanjit

On 8/29/07, Reza Lotun <rlotun at cs.ubc.ca> wrote:
>
> Hi,
>
> Thanks for the replies. I suppose, for a concrete topology I'd like to
> have 5-hop configuration consisting of 4 AP nodes, each with 1 node
> connected to each ap. i'd like to make it so that some AP-node pairs
> are completely out of range of each other, and some nodes should
> interfere in a hidden-sender and exposed terminal situations.
>
> I suppose the question is, can topogies be set up so that nodes are
> not only out of range, but in interference range (i.e. can't decode
> transmissions from other nodes, but those transmissions still
> interfere with it's own transmissions)?
>
> Thanks!
> Reza
>
> On 8/24/07, Sanjit Kaul <sanjitkaul at gmail.com> wrote:
> > Hi Reza,
> >
> > We do have an algorithm to map topologies on to the ORBIT grid, using
> noise
> > injection. The algorithm has not been released to public yet.
> >
> > I request you to send out the details of the topology you are interested
> in.
> > Currently we treat any topology (infrastructure/ ad-hoc) as a bunch of
> up or
> > down links (connected / dis-connected).
> >
> > Given the topology I could tell you how feasible it is to map on the
> grid
> > and the caveats.
> >
> > Lets not go by the word arbitrary, please :).
> >
> > Regards,
> > Sanjit
> >
> >
> > On 8/24/07, Ivan Seskar <Seskar at winlab.rutgers.edu> wrote:
> > > Hi Reza,
> > >
> > > I will let others respond to your particular topology question, but
> just
> > > want to comment on the possibility of triggering carrier sense with
> > > noise injection: it is certainly possible to increase noise level to a
> > > point of triggering energy detection threshold (mode 1) in the cards
> > > (especially for the nodes that are in the vicinity of injection
> > > antennas).
> > >
> > > Regards,
> > >
> > > Ivan.
> > >
> > > PS; In addition to noise levels, you should also take into
> consideration
> > > transmit power and CS threshold as knobs that can be used for topology
> > > creation.
> > >
> > > -----Original Message-----
> > > From: owner-orbit-user at winlab.rutgers.edu
> > > [mailto:owner-orbit-user at winlab.rutgers.edu] On Behalf Of
> > Reza Lotun
> > > Sent: Friday, August 24, 2007 5:16 PM
> > > To: orbit-user at winlab.rutgers.edu
> > > Subject: ORBIT-USER: Setting up Topologies
> > >
> > > Hello,
> > >
> > > I'd like to pose a general question to the list. Has anyone attempted
> > > (and successfully) deployed arbitrary topologies on the grid.
> > >
> > > A clarification. According to the orbit wiki, the Orbit approach to
> > > creating arbitrary (multi-hop) topologies is to use noise injection:
> > > http://www.orbit-lab.org/wiki/HowTo/UseNoise
> > > As far as I can see, the idea is to inject constant background white
> > > gaussian noise to raise the noise level, so that only more powerful
> > > signals can be discerned.
> > >
> > > I'm interested in simulating a real wireless network deployed in a
> large
> > > space operating in convential infrastructure mode - this means that
> some
> > > nodes will naturally be out of range of one another. Is there ever a
> > > danger when injecting noise that the nodes will naturally back-off
> since
> > > they sense transmission? Is anyone using orbit to test non-mesh-like
> > > 802.11 topologies, that is, the tree topology of
> > > client->access_point common to infrastructure networks?
> > >
> > > Thanks!
> > > Reza
> > >
> > >
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://orbit-lab.org/pipermail/orbit-user/attachments/20070830/6f07a74e/attachment-0001.htm