Point coordinator functions / HostAP

[Dave Edwards]

I've been working on something similar to this. However, the problem 
here is the radio themselves,. A polling protocol requires that the 
stations have a short interframe space between being polled and 
generating the ACK/Data. As I've said before, the Prism 2 PCCard 
solution is just not fast enough on its IO. Perhaps the prism 3 chipset 
is better, but I don't know!

The solution is to use a DMA based PCI or Cardbus chipset, however 
Intersil do not make such a thing, which means that we would have to use 
one of the other Vendors. Given the current situation with lack of 
information about the other chipsets and NDA clauses preventing the 
opening of source code, I'd have to conclude that an opensource polling 
algorithm will not happen.

Dave



Brian Capouch wrote:

> Dave Edwards wrote:>
>
>>
>> My apologies for being a PITA but I've spent the past few months 
>> looking at this
>>
>
> I wonder about an alternative, such as is done with Karlnet's 
> Turbocell, that instead of using 802.11b uses a proprietary polling 
> protocol riding on top of standard independent-mode point-to-point 
> associations.
>
> I haven't spent much time digging into code, but I've spent a ton of 
> time talking to lots of smart people about this, and (unless I'm 
> missing something, which certainly could be possible) it would seem 
> like the undertaking of an open source project to add polling 
> functionality to standard ad-hoc links would have a huge payoff.
>
> Karlnet's software can do polling with standard omni antennas, and 
> from what I can tell their claim of high throughput and distance 
> insensitivity are both substantiated in practice.
>
> I would love to see this discussed by people who are more MAC- and 
> PHY-layer literate than I am.
>
> Thanks.
>
> B.
>
>
>
>