[OLSR-users] Double radio mesh test
Mon Apr 18 20:51:35 CEST 2005
Stefan Sayer <(spam-protected)> :
> I wonder whether meshdynamics' multi radio network is configured in a
> distributed way or whether there is e.g. a central station that would
> let the other station channel hop - they say that the network will
> dynamically adapt to interference.
> If it is really distributedly self configuring then I wonder how a
> station can find a neighbor station that has hopped to another channel
> to talk to another node (avoiding interference on the previous channel)
> without scanning all the time. consider this:
> first A is talking to B on ch1
> A 1 ----------- 1 B
> C D
> then A and C hop to ch 6 while e.g. B talks to D on ch1
> A 6 1 B
> | |
> C 6 1 D
> then B wants to talk to C: how does it know on which channel ?
(I assume that B does not see C directly)
Then you have 2 options:
A 1--------------1 B
A 6--------------6 B
The protocol should choose solution 1 or 2 by having either a "Link
Quality" value for each connexion, or either a direct bandwidth
measurement (which can vary with the traffic). In this last case, if the
traffic is high on one connexion at time t, it won't be necessary the
case at t+1. Maybe a value of "real available bandwidth" would be
interesting to have for the protocol, as a mean to choose the best path.
> In their graphics about 2,3,4 radios they never have a more realistic
> mesh situation where some nodes are in the same range - there is alway
> only two nodes in one circle.
Yes, that's a problem. Choosing which neighboor to transmit is not a
problem, since the node has not the choice in this case.
The problem is to get values of throughput for each link. No?
> Jeromie Reeves wrote:
> >That should work fine. The trick will be to use manual channel selection
> >to avoid having the same channel space
> >on every single unit (IE 1 & 11 on every unit).
> >1 & 11 --- 1 & 6 --- 6 & 11
> >That way you force the packets to change radios and help stop self
> >There is no problem having units around that also use the same channels,
> >just dont
> >do it network wide.
> >Henrion Benjamin wrote:
> >>Very interesting report here:
> >>You only need 2 radios/node if you don't care about "people with their
> >>laptops" who access with the 3rd.
> >>Is it possible to do the same with OLSR? Maybe OLSR running on 2
> >>different wireless interfaces on 2 different channels (like 1 and 11)?
> >>Do you think it might be possible?
> >>Benjamin Henrion <(spam-protected)>
> >><< Software patents are a
> >>Temptation >>>
> >><< Temptation leads to
> >>Stagnation >>>
> >><< Stagnation leads to the Dark
> >>Side. >>>
> >>olsr-users mailing list
> >olsr-users mailing list
> olsr-users mailing list
Benjamin Henrion <(spam-protected)>
<< Software patents are a Temptation >>>
<< Temptation leads to Stagnation >>>
<< Stagnation leads to the Dark Side. >>>
More information about the Olsr-users