Re: prism2 silence value question

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tue, 28 Jan 2003, h1kari wrote:

> Rob,
>
> I've had a similar experience with some of my cards.. My older prism2 cards
> (smc 2632w v8r3) and my senao 200mw prism2.5 card will not report any
> (accurate) noise values. Whereas, my milan prism2.5 card (under the sticker
> it says prism 2.5 reference card) reports accurate noise values. I'm
> guessing that it might be something that intersil forgot to implement
> originally and finally got around to implementing it properly.
>
Hrmmm...you think it may be a prism2 vs. prism2.5 issue? That seems
reasonablem but doesn't really apply to what I'm seeing since the cards
(supposedly) have the same chipset/firmware. ;) Also, how are you defining
"accurate noise values"? I'm guessing that you're referring to the > 27
values as the rest get thrown in the bitbucket in your code. I have to
say that in my experience with my Netgear card, even the packets w/ noise
< 27 are still quite useful in determining signal and noise info. The
only real values that have been in the least troublesome to me are the
values that are completely out of whack in relation to the rest of the
values I'm getting.  For example, if I get packets with sig/noise values
that looks like:

 pkt  sig  noise
 ---------------
  1   83    6
  2   90    8
  3   87    7
  4   63    6
  5   86    7

The only packet that tends to skew my results is (obviously) packet 4.
Forunately a little simple math helps to smooth the values out. I'm
currently using averaging across all packets seen from a station to do
that, but I've also given thought to a sliding window of packets and a
little standard deviation math. Averaging is good enough for now, so I'm
sticking with it. In case you haven't guessed, I'm messing with ideas for
site survey stuff :)

> Additionally, I remember reading somewhere that the signal/noise/quality
> values are calculated based off of the ratio of good packets to crc err
> packets. If that's actually true, it might be possible to implement noise
> ratings on the software end. Just a thought.
>
That's interesting. You wouldn't happen to have a link to that info buried
anyplace would you? If I could determine how they're calculating the
sig/noise values, I'd happily do it myself. As it stands, with a quick
think about it, I can't imagine how they'd derive sig/noise values for
packets from a specific station off of the CRC values. I mean, who's to
say that the addr fields didn't get corrupted?


- -d

"You see, wire telegraph is a kind of a very, very long cat.
 You pull his tail in New York and his head is meowing in
 Los Angeles. Do you understand this? And radio operates
 exactly the same way: you send signals here, they receive
 them there. The only difference is that there is no cat. "

                                           --Albert Einstein

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE+NxLsiIu2X5vnl3ERAhL4AJ9Wn902hXukFHRG5IFWv68EA03FlACgp75X
pGMIwR5C5vqQ6e0OTkyDBDQ=
=okl8
-----END PGP SIGNATURE-----