The question I have is what do mean by 'break' the connection?  We have
turned off a node and it took about 20 seconds to re-route.  We can try the
'plink_action block' next and see how long it takes to re-route.

Y-

On Mon, Apr 23, 2012 at 10:58 AM, Javier Cardona <[email protected]> wrote:

> Hi David,
>
> On Mon, Apr 23, 2012 at 9:37 AM, David Fulgham
> <[email protected]> wrote:
> > We've been playing with the idea of a mobile 802.11s mesh for a while,
> > since we have a mesh through out an entire community we would like to
> > be able to use it for voice, etc. The problem is that when a node
> > moves through the network it leaves remnant of itself throughout the
> > network.  I.e. nodes will attempt to use the mobile node if it
> > happened to result in the best path to a given destination at some
> > point in time.  We found that the timeouts for path eliminations even
> > when we reduced the timeouts were not responding to the changes fast
> > enough.
>
> Establishing a peer link is costlier than establishing a path through
> the mesh (specially if security is used).  For that reason, peer links
> stay around longer, while paths change more rapidly in response to
> topology changes.  So the fact that you see peer links long after a
> particular peer has moved away is not necessarily a problem.
>
> To test mobility, I would recommend using the simulated device driver
> mac80211_hwsim.  With it you can make and break topologies in real
> time and see how the 802.11s protocol responds.
> For instance, one of our regression tests is as follows:
>
> 1. Establish a topology like:
>
> # 1 - 2... n-2 - n-1
> # \------n------/
>
> 2. Ping from 1 to n-1, confirm that traffic flows through n
> 3. Disconnect n from the network
> 4. Measure how long until traffic is re-routed through 2-..-n-2
>
> With n=5 we observe a path recovery in less than 0.7 sec.
>
> Cheers,
>
> Javier
>
>
> >
> > On Mon, Apr 23, 2012 at 12:29 PM, Yile Ku <[email protected]> wrote:
> >> I am looking for someone who can answer questions related to:
> >> 802.11s mobility.  Nodes moving relative to one another, how fast routes
> >> across the mesh converge
> >> Power control, how to get the maximum amount of power from the cards we
> are
> >> using (Mikrotek)
> >> Parameter configuration for mesh networks.
> >>
> >> Thx,
> >> Y-
> >>
> >>
> >> On Mon, Apr 23, 2012 at 9:45 AM, David Fulgham <
> [email protected]>
> >> wrote:
> >>>
> >>> We've been deploying 802.11s in a "wide area" network covering roughly
> >>> 3-4km.sq. and would be available to consult if required.
> >>>
> >>> David Fulgham
> >>> Mnemonics.ca
> >>> www.mnemonics.ca
> >>>
> >>> On Mon, Apr 23, 2012 at 11:00 AM, Yile Ku <[email protected]>
> wrote:
> >>> > Anyone besides Cozybit do consulting for 802.11s?  I am interested
> in a
> >>> > guru
> >>> > since we need some questions answered quickly and thoroughly.
> >>> > Y-
> >>> >
> >>> > _______________________________________________
> >>> > Devel mailing list
> >>> > [email protected]
> >>> > http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
> >>> >
> >>> _______________________________________________
> >>> Devel mailing list
> >>> [email protected]
> >>> http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
> >>
> >>
> >>
> >> _______________________________________________
> >> Devel mailing list
> >> [email protected]
> >> http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
> >>
> > _______________________________________________
> > Devel mailing list
> > [email protected]
> > http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
>
>
>
> --
> Javier Cardona
> cozybit Inc.
> http://www.cozybit.com
> _______________________________________________
> Devel mailing list
> [email protected]
> http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
>
_______________________________________________
Devel mailing list
[email protected]
http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel

Reply via email to