Re: [Madwifi-devel] [Fwd: Disable 11a turbo mode]

[Sebastian R. <Seb...@uc...>]

Hi Benoit,

Thanks for your reply. Comments in line...


Rgds,
Sebastian

Benoit PAPILLAULT wrote:
> Sebastian Robitzsch a écrit :
>> Hi again,
> 
>> Just wondering whether no one knows anything about this issue?!
> 
>> BTW - I'm aware of this page:
>> http://madwifi-project.org/wiki/UserDocs/TurboMode
> 
>> However, just the answer "don't know" does not really satisfy me ;-)
> 
> 
>> Rgds,
>> Sebastian
> 
>> -------- Original Message --------
>> Subject: [Madwifi-devel] Disable 11a turbo mode
>> Date: Wed, 03 Jun 2009 12:31:36 +0100
>> From: Sebastian Robitzsch <Seb...@uc...>
>> Organization: University College Dublin
>> To: mad...@li...
> 
>> Hi folks,
> 
>> I've performed comprehensive measurements for .11a. Before explaining my
>> problem in detail here are my configurations/equipment:
> 
>> - Kernel 2.6.26 (Debian 5)
>> - MadWiFi 0.9.4 Rev 4011
>> - Routerboard 802.11a/b/g miniPCI card with AR5414 chipset
>> - TSHARK 1.0.2
> 
>> After using a spectrum analyser to investigate the adjacent channel
>> interference problem I got unpredictable peaks over the entire .11a
>> frequency space. After obtaining the DataRate out of the radio-tap
>> header of the received package by using Tshark I saw _JUST_ Beacons sent
>> by using a datarate of 12 Mbps. Further, the TSHARK log showed only user
>> data rates above 54 Mbps (mainly 108) for the user data transmission
>> (using IPERF tool).
> 
>> I tried to disable the Turbo mode by:
>> # iwpriv <interface> mode 1
> 
>> or
>> # iwpriv <interface> turbo 0
> 
>> or
>> # iwpriv <interface> 11a
> 
>> or
>> # iwpriv <interface> 11a fixed
> 
>> Additionally, I tried to change the countrycode to Ireland, Germany or
>> wildcard. At any time, I tried the following to look whether something
>> has changed:
>> # iwpriv <interface> get_mode
> 
>> gives _ALWAYS_:
>> # <interface> get_mode:11a
> 
>> Using
>> # ath_info <HW ADDRESS>
> 
>> Gives also always:
>> | 802.11a Support:    yes | Turbo-A disabled: no |
> 
> 
>> Further, I tested infrastructure mode as well as adhoc mode. However, I
>> still see JUST beacons sent in turbo mode (12Mbps) ... after playing
>> around with kernel and madwifi versions (0.9.3 - rev4011) and no reply
>> on the madwifi user mailing list, I'm writing now to this mailing list.
> 
>> Any suggestions?
> 
> 
>> Sebastian
> 
>> ------------------------------------------------------------------------------
>> Crystal Reports - New Free Runtime and 30 Day Trial
>> Check out the new simplified licensing option that enables unlimited
>> royalty-free distribution of the report engine for externally facing
>> server and web deployment.
>> http://p.sf.net/sfu/businessobjects
> 
>> ----------------------------------------------------------------------
> 
>> _______________________________________________
>> Madwifi-devel mailing list
>> Mad...@li...
>> https://lists.sourceforge.net/lists/listinfo/madwifi-devel
> Hi Sebastian,
> 
> I'm pretty sure you are facing two problems:
> 
> - what you think is turbo mode might not be turbo mode. AFAIK, turbo
> mode is not related to data rate. Turbo mode is linked to using 40 MHz
> of frequency bandwidth rather than 20 MHz (it's OFDM, so I think the
> data rate is linked to the radio modulation over each sub carrier X
> coding rate X  number of  data sub carrier.  Right?).

Sorry... but you are wrong. I'm talking about "Dynamic Turbo", i.e., two
20MHz bandwidth radio channels. And finally you are getting 108 Mbps
nominal datarate.
http://www.super-g.com/collateral/atheros_superg_whitepaper.pdf (Page 12)

> Could you
> provide a capture file done with tshark for instance?

No need to give you the tshark logfile. I used the following command
# tshark -i mon0 -n -T fields -e radiotap.datarate

the result (without sending user data) was just:
12
12
12
12
12
12
12
...


with user traffic just:
108
108
108
108
108
108
...

> 
> 
> - If you got peak over the entire .11a band, that might be linked to
> using rate with a 0 value which, according to other message on this
> list, will generate a lot of noise... So you need to check this point
> as well.

Please have a look to the attached gnuplot picture. The guy on channel
36 was sending IPERF generated traffic. That is more than just noise -
this is definitely turbo mode.

> 
> I don't have a spectrum analyzer at hand, otherwise, I would have
> check those 2 points myself....
> 
> Regards,
> Benoît
> 

-- 
Mr. Sebastian Robitzsch, Dipl.-Ing. (FH)
PhD Candidate
University College Dublin
School of Computer Science & Informatics
Belfield
Dublin 4
Ireland


Mobile:		+353 (0)86 - 22 10 988 (IRL)
		+49 (0)1 72 - 3 52 80 72 (D)
Mail:		seb...@uc...
WWW:		www.seronline.de

Skype:		seronline82
Jabber:		ser...@am...

Delicious:	www.delicious.com/seronline
Bibsonomy:	www.bibsonomy.org/user/seronline
-----------------------------------------------------