Hi,

 I still wonder if setting up a static BSSID will stop the BSSID

selection algorithm from ocurring. What do you think? Have tested this in practice?

yes, it will deactivate the selection algorithm. We used this hack to get the Freifunk Berlin network stable. We had a real hardware zoo (Atheros, Broadcom, Ralink, etc) that was impossible to control without setting the BSSID.

Regards, Marek

Hi,

Have you managed to set a static BSSID to all those chipsets - Atheros, Broadcom, Ralink etc? Because up to now, only Atheros made this trick possible.

thanks to Felix from the OpenWRT project the Atheros driver has that feature (as you know). Broadcom is a much harder cases because the drivers a binary only. We mostly use the famous WRT54G and its derivates for our network. Sven- Ola managed to write patch for that binary driver of the WRT. There also is an open source driver project for the broadcom chips but I'm not sure where it stands. I never experimented with Ralink myself, so I can't answer that. I'm sure Elektra could jump in here. :-)

Regards, Marek

Hi -

I'm waiting for working drivers (particularly USB) for IBSS mode for long time and was putting my hope on Ralink RT73 as well.

The situation is improving with the mac80211 driver development in the Linux kernel. Recently patches went into compat-wireless that allow to fix the IBSSID with the

iwconfig <interface> ap <cell-id>

command - with some drivers, not all.

However the situation is a little more complex than just fixing the IBSSID, because the drivers/firmware have to be convinced not to perform a IBSS-merge every time the TSF timestamp in a received beacon is a little older. Upon a merge the cards may flush their MAC table - this could happen every few miliseconds - like I have observed with Intel ipw4965 driver. Also TSF clock skews can cause race conditions.

So far ath5k and ath9k seem to work - but the only driver I can really recommend is Madwifi with all patches from Openwrt - vanilla Madwifi is *not* working. The patched Madwifi in Openwrt Kamikaze works perfectly (I'm using mesh with Batman exclusively every day and I can suspend/resume my EEE-PC for weeks and I'm always online ;-)

But Madwifi does not support USB...

Best bet for USB is Zydas, I guess.

cu elektra

Hi,

 Thanks for the information. At least now I have some background

to work on - I already have several rt73 USB dongles, some notebooks with Atheros, some OpenMokos (which have atheros chipsets) and a few Nokia N810 tablets (which have a prism chipset).

although the Openmoko Freerunner has an Atheros chip inside it does not use the madiwifi driver (this is the AR6001 mobile chip, in case you wonder).

 And they mentioned about some intermediate mode, called

Pseudo-IBSS or AHdemo mode, which allows a device to set a static BSSID and it is supported by some drivers. I suppose we are talking about the same thing here, just not giving any names (?).

AHdemo is not the same as setting the BSSID in ad-hoc mode manually. As mentioned on the page you linked to the AHdemo mode does not send any beacons. Whereas the ad-hoc mode sends beacons with a fixed BSSID. Not sending beacons has certain side effects: - your network will appear being "invisible" (the normal network manager wont show it) - you have to manually configure the connection speed and other stuff as the usual autonegotiation is using the beacons for that - you will have collisions (might lead to decreased performance) - driver compat problems (only a few drivers support this mode)

Nevertheless, it will fix your cell split problem. :-)

 When you mentioned your EEE PC, which uses an Atheros chipset,

did you also had to set a static BSSID to the same used in the mesh network you use?

I think EEE PC uses the madwifi driver?! If so you can set the BSSID statically.

Regards, Marek

Hello Marek, Elektra,

Thank you very much for the information.

By the way, I think you guys are involved on that book project about community wireless networks, hosted at WNDW.net. I'm from Brazil, and I have noticed that the only translation available is to the first edition, but the book seems to be on its thrid edition actually.

I'd be interested to cooperate on the translation of the book, as I intend to use them on my classes and also I intend to start some community network projects where I live. Do you know who should I contact in order to help on the translation?

regards,

2009/2/8 elektra onelektra@gmx.net:

Hi -

some EEE PC models use a Atheros mPCI wireless card (the 701 for example) but my 901 was shipped with a Ralink rt2870 (802.11 abgn) that I replaced with an Atheros abg card ASAP...

Actually I'm using Ahdemo mode in my EEE-PC which works nicely together with other devices with 'normal' IBSS-mode but fixed IBSS-ID - in our local mesh we use Madwifi with OpenWRT patches and Broadcom-based devices (WRT54G/GL/GS, Asus WL500GP) with OpenWRT Whiterussian and Broadcom proprietary driver with fixed IBSS-ID hack (Freifunk Firmware).

Rate adaption with the Ministrel algorithm works well in Ahdemo mode - and you have the possibility to set up a AP VAP on the same interface, too.

Performance is fine. That may be different if you run a Ahdemo-only network in range with other wireless networks - they won't detect collisions, hence performance can suffer.

Cheers, elektra

---

B.A.T.M.A.N mailing list B.A.T.M.A.N@open-mesh.net https://lists.open-mesh.net/mm/listinfo/b.a.t.m.a.n

Hi,

How about in the laptops / pcs that access the WRT54G, which make

the "infrastructure" of the mesh? What kind of adapters do you use on them?

I think Elektra made clear that the driver situation for mobile devices is much worse. We stopped promoting laptops as part of the network. Next to the driver issues we faced different OSes in different versions, firewalls, $security_software, wrongly configured routing daemons, confused ordinary users, etc. Instead we encouraged people to use the routers for meshing and normal managed networks for the laptops.

Regards, Marek