On Sunday afternoon, we finally managed to execute a number of test scenarios in the 10-nodes setup. Nodes were places between a horse stable, along a street and the house hosting the WBMv3.3 event. The following diagram illustrate the setup and the existing links:

     |    ch01     |     ch01    |
 9---+             |             |
     |             |             |
          ch13           ch13    :     ch13      :
 house           pole           car             horse stand

...  :  24BMit wireless link using ch01 or ch13
---  |  100MBit ethernet link

There was line of sight (LoS) between the locations:

* house - pole
* pole - car
* car - horse stand

Transmit power of all nodes was fixed to 5dBm

Data rate of all nodes was fixed to 24MBit

The following network ranges were used:

* 10.10.x00.Node/24   static link local
* 10.10.x01.Node/24   batman  (udp.port == 4305)
* 10.10.x02.Node/24   babelz  (udp.port == 8475)
* 10.10.x02.Node/24   olsr    (udp.port == 698)
* 10.10.x02.Node/24   bmx6    (udp.port == 6240)

where x could be:

* x=2 for primary IP assigned to lan interface of node
* x=1 for non-primary IP assigned to wireless interface of node

Node 9 was a notebook participating in the mesh via ethernet and running all 4 routing protocols.

Test scenarios were initiated from node 10.

From 10 we send simultaneous ping recored-route probes to 9 and 11 to observe the selected routes chosen by different protocols.

In parallel, we also downloaded data from node 9 to node 10 to saturate the links.

The test were executed by opening several ssh-shells into node 10 from the notebook. Logfiles of the different shells can be downloaded here:







At some point, we have a very stable link between the Horse and the 9 laptop, with speed of about 350KO/sec to 400KO/sec.

BattleMeshV3.3/HorseTests (last edited 2010-11-16 17:16:09 by zoobab)