2008-01-10 12:14:50

by Johannes Berg

[permalink] [raw]
Subject: channel utilisation short/long hdr?

What's the meaning of these defines?

#define CHAN_UTIL_HDR_LONG (202 * CHAN_UTIL_PER_USEC)
#define CHAN_UTIL_HDR_SHORT (40 * CHAN_UTIL_PER_USEC)

I can't figure out what the values 202 and 40 are.

And probably related, what does it mean for a bitrate to be an "ERP
rate"? Does that just mean it's an OFDM bitrate?

johannes


Attachments:
signature.asc (828.00 B)
This is a digitally signed message part

2008-01-12 02:45:21

by Jouni Malinen

[permalink] [raw]
Subject: Re: channel utilisation short/long hdr?

On Thu, Jan 10, 2008 at 12:50:58AM +0100, Johannes Berg wrote:
> What's the meaning of these defines?
>
> #define CHAN_UTIL_HDR_LONG (202 * CHAN_UTIL_PER_USEC)
> #define CHAN_UTIL_HDR_SHORT (40 * CHAN_UTIL_PER_USEC)
>
> I can't figure out what the values 202 and 40 are.

Hmm.. I thought I had filtered out the channel use estimation code from
the original release since I did not see much use for this and the
values were not exactly valid in many cases.. ;-).

This code is almost five years old and maybe not so surprisingly, I
don't remember the exact details anymore.. Since these are use as
"hdrtime", I would assume they were derived using following calculation:

CHAN_UTIL_HDR_LONG: assume 802.11b with long preamble:
aSIFSTime (10 usec) + aPreambleLength (144 usec) + aPLCPHeaderLength (48
usec) = 202 usec

CHAN_UTIL_HDR_SHORT: assume OFDM:

aSIFSTime (+ signal ext in case of 802.11g; 16 usec) + preamble (16
usec) + signal (4 usec) = 36 usec; add one symbol time (4 usec) for
some reason (or a sign of someone not knowing how to add small
numbers together.. ;-), I don't remember anymore and cannot figure out
this without wasting way too much time..)

> And probably related, what does it mean for a bitrate to be an "ERP
> rate"? Does that just mean it's an OFDM bitrate?

That would be any rate using Extended Rate PHY (ERP; 802.11 Clause 19).
This is not necessarily OFDM (it could also be using PBCC modulation).
In practice, I would expect the kernel side implementation to use this
in anything PHY related as ERP-OFDM (i.e., it may not support optional
PBCC rates). The association and beacon/probe processing may be more
generic to any ERP rate. I don't think ERP-PBCC is very commonly used,
so this may be more on the theoretical side..

--
Jouni Malinen PGP id EFC895FA

2008-01-12 10:15:17

by Johannes Berg

[permalink] [raw]
Subject: Re: channel utilisation short/long hdr?


> Hmm.. I thought I had filtered out the channel use estimation code from
> the original release since I did not see much use for this and the
> values were not exactly valid in many cases.. ;-).

Heh.

> This code is almost five years old and maybe not so surprisingly, I
> don't remember the exact details anymore.. Since these are use as
> "hdrtime", I would assume they were derived using following calculation:

Right. Thanks for the values, I guess that if we really wanted to keep
that stuff maintainable we should rewrite it and take into account
preamble length too etc. I'm not too keen on that though, would anybody
object if we'd just kill it?

> That would be any rate using Extended Rate PHY (ERP; 802.11 Clause 19).
> This is not necessarily OFDM (it could also be using PBCC modulation).
> In practice, I would expect the kernel side implementation to use this
> in anything PHY related as ERP-OFDM (i.e., it may not support optional
> PBCC rates). The association and beacon/probe processing may be more
> generic to any ERP rate. I don't think ERP-PBCC is very commonly used,
> so this may be more on the theoretical side..

Oh, right, I keep forgetting about the optional rates.

Thanks,
johannes


Attachments:
signature.asc (828.00 B)
This is a digitally signed message part