Hi everyone,
I'm new to the list, although I've been using batman-adv for a year now via a couple of networks I've deployed using Mesh Potatoes (based on openwrt). Thank you very much for the work, the result is very, very good.
In one of the networks I'm facing from the beginning some problems due to collisions, thus reducing the TQ in some links, that I'd like to avoid in order to optimize performance. One of the messages that could be colliding with the OGMs are the WiFi beacons. I'm a bit confused about the beacon transmission. I posted these questions in the villagetelco list, but they recommended me to ask them here too. Hope you can help me.
The mesh is created using ahdemo and according to Madwifi this mode does not use beacons. However, athstats is showing many beacons being transmitted. I guess those are the ones transmitted from the VAP also configured in the same physical interface as the ahdemo. The option sw_merge is recommended to be active in the ahdemo interface to avoid the hardware to try to synchronize the timestamps from the beacons. In addition, it is recommended to create the VAP with option nosbeacon activated for the same reason.
According to athstats, beacons are being tx, I guess those from the VAP. However, I'm not sure if athstats is reporting properly, since the chipset does not have diversity for TX, but athstats is reporting so. - If there are no beacons in the mesh (due to ahdemo), or even if there are (from the VAP), and we disable the synchronization of timestamps with the options sw_merge and nosbeacon, how the MPs get their time synchronization as for sharing the same reference when using protocol timers (slottimes, DIFS, etc...) so important for avoiding collisions? - If there are no beacons in the mesh, how do the nodes get the information contained in the beacons about the capabilities of their neighbors (protection, short preamble, etc) as for allowing 11bg seamless operation, from the VAP beacons? - Given the network it is going to be, at least during some months, very static (both in mesh nodes and clients) and assuming the beacons are coming from the VAP, what values would make sense to use as beacon interval and OGM interval, respectively, for avoiding collisions and keeping performance?
Thank you very much in advance,
Regards
carlos
-- Note that I am using a new email account carlos.reymoreno@gmail.com, and that the former email address crey@ehas.org will be disappearing soon. Thanks for changing this in your contact list.
Carlos Rey-Moreno Research Assistant Office 1.28 Department of Computer Science University of the Western Cape Private Bag X17 - Bellville, 7535 Cape Town - South Africa
Tel: +27 (0) 21 959 2562 Cel: +27 (0) 76 986 3633
Skype: carlos.reymoreno Twitter: Creym
On Tuesday, February 12, 2013 13:00:33 Carlos Rey-Moreno wrote:
In one of the networks I'm facing from the beginning some problems due to collisions, thus reducing the TQ in some links, that I'd like to avoid in order to optimize performance. One of the messages that could be colliding with the OGMs are the WiFi beacons. I'm a bit confused about the beacon transmission. I posted these questions in the villagetelco list, but they recommended me to ask them here too. Hope you can help me.
Don't know who sent you here but this list does not seem like the right spot for your questions. Most of them center around madwifi, ahdemo mode and beaconing - none of these are related to batman-adv (unless I misunderstand?).
- Given the network it is going to be, at least during some months,
very static (both in mesh nodes and clients) and assuming the beacons are coming from the VAP, what values would make sense to use as beacon interval and OGM interval, respectively, for avoiding collisions and keeping performance?
If you wish to avoid packet collisions in the air I suggest sending less packets, hence make the intervals longer. Suggesting exact values is a little hard as those always depend on your environment & setup. I'd recommend to experiment with these values to find out what works for you.
Cheers, Marek
b.a.t.m.a.n@lists.open-mesh.org