Hi David,

On Mon, Apr 29, 2013 at 2:32 PM, David Romero <[email protected]> wrote:
> I've been testing the implementation for several months with many tests, and
> I tell you some of my strange results that I've checked repeated, may be you
> have an answer to them, or a solution.
>
> 1: Two nodes, in the same room, working in a quiet radio environment. I
> sniff all the traffic with a laptop and kismet. I noticed many times in the
> connection establishment phase, that a node A sends to another node, three or
> four times "Mesh peering open", and the other node answers with an ack to
> each one MPO, later, node A sends a "The mesh STA has resent dot11MeshMax
> retries Mesh Peering messages...without a mesh peering confirm", 100ms later
> node B send a  "Mesh peering open", an both nodes establish the connection
> properly. ¿Why node B does't answer to the first 4 messages?

Can't think of a reason.  Maybe we can take a look at your capture file.

> 2 Another strange behavior that is repeated many times in my tests. A
> network with 4 nodes in line  A-B-C-D  (A only has a B peer..., B has A and
> C as peers...), working in UDP, I check firstly from node A to the rest of
> nodes that BW is normal, with max of 15 Mbps.
>
> I send udp iperf of 5 Mbps from A to C without any problem checking with a 3
> seconds interval., at the moment I start a simultaneous new transfer form B
> to D of 5 Mbps, the first transfer decrease to 20Kbps, and transfer from B
> to D goes without any problem. At the moment that transfer form B to D
> stops,the bad trasfer from A to C goes up to almost 5 Mbps. I've changed the
> HW, the machines, the OS, every time is the same.
>
> In another test, one trasfer A to D and another B to C, the low BW affects
> to A-D trasfer. ¿any idea? ¿SW problem?

All i can think is that if all nodes are within range, they may be
contending for airtime.
Can that be the reason?

Cheers,

Javier

-- 
Javier Cardona
cozybit Inc.
http://www.cozybit.com
_______________________________________________
Devel mailing list
[email protected]
http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel

Reply via email to