Maximizing 802.11b throuput

[Bichot Guillaume Princeton]

> 
> You also need to factor in the card's io performance. For the pccard 
> based prism2 devices this can be as long as 200uS to write a 
> frame into 
> the device. I'm not sure if it's an issue with the PCI based 
> devices but 
> be aware, the pccard (16bit) interface is painfully slow!

To be exact it should add delay but should not reduce the throughput (16bits
PCMCIA = 20 Mbytes/s). 

> 
> This can be a problem if you are implementing a polling protocol with 
> the radios.
> 
> Also, some versions of the Prism 2 firmware will only send multicast 
> frames after a beacon. The AP firmware will allow multicast 
> frames to be 
> sent at other times, but it's difficult getting a legal copy of it!

That is true. This implies big limitatino as I have constated with my card.

> 
> 
> Bichot Guillaume (Princeton) wrote:
> 
> > In theory the overal bit rate would be not so far from 11 Mbits. If
> > I'm not wrong, based on payload size of 1500 bytes, a total 
> overhead 
> > of 1148 bytes max (including PHY (guard interval + 1Mbit/s 
> overhead) , 
> > MAC (DFIS + data frame header), LLC header, SNAP header , 
> IP header, 
> > UDP header)  would give you a total throughput of 7.8 Mbit/s.
> > This is for one way streaming in multicast/broadcast without any 
> > IEEE802.11 association attempt (no uplink traffic at all) and good 
> > channel quality.
> >  
> > In multicast however I've constated that the bit rate is sometime
> > limited by the firmware of my card (PCI prism2 Linksys). I cannot 
> > stream more than 1Mbit/s. 
> >  
> > If you stream in unicast this is another story. MAC 
> acknowledgment may
> > reduce considerably the bandwidth depending on the quality of your 
> > channel. Practically the 3Mbit/s number seems often raised.
> >  
> > In theory transmitting in multicast with the maximum packet size is
> > the way to maximize the bandwidth usage. However you have 
> to cope in 
> > some way with error correction.
> >  
> > Guillaume Bichot
> > 
> ----------------------------------------------------------------------
> > ----------
> > As of February 12th, 2003 Thomson unified its email addresses on a 
> > worldwide basis. Please note my new email address: 
> > guillaume.bichot at thomson.net http://www.thomson.net/
> >
> >     
----------------------------------------------------------------------------
---- 
As of February 12th, 2003 Thomson unified its email addresses on a worldwide
basis. 
Please note my new email address: guillaume.bichot at thomson.net 

http://www.thomson.net/ 

----Original Message-----
> >     From: Andr? Luiz Ribeiro Moutinho
> >     [mailto:andre.moutinho at compsisnet.com.br]
> >     Sent: Thursday, March 13, 2003 8:25 AM
> >     To: hostap at shmoo.com
> >     Cc: Mauricio Micoski; Marcelo Ferreira Vinhas; Marco 
> Ant?nio Fernandes
> >     Subject: Maximizing 802.11b throuput
> >
> >     Hi,
> >
> >     I am implementing a realtime video streamming application using
> >     802.11b in order to transmit video data.
> >     Transmitting MJPEG 320x240 at 30fps and 40% quality results a
> >     3Mbit/s total bandwith. When I try to increase the video data
> >     bandwith (for instance, rising JPEG quality), the radio channel
> >     starts
> >     failing to transmit all video information. I would like 
> to know what
> >     is the REAL 802.11 data transmittion throuput (4 or 
> 5Mbit/s ??) and
> >     what could be done in order to maximize the data throuput. For
> >     instance,
> >     trying to optimize the transmitted data block size.
> >     I need to use the maximum data bandwith available in order to
> >     transmit
> >     the maximum mount of video sessions possible.
> >
> >     Thanks a lot,
> >
> >     Andre Moutinho
> >
> 
> 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.shmoo.com/pipermail/hostap/attachments/20030313/65ab8e49/attachment.htm