[ath5k-devel] Combined rate and antenna selection algorithm

2012-03-15 Thread jpo 
Hello all,
I'd like to present and maybe get a little bit feedback from the community for a
rather unique problem we face in a production environment. We are using EMP-8602
PLUS-S cards in 802.11a mode (5GHz/BBDR) with an external TX amplifier attached
to one of the antenna ports. The second antenna port is not amplified. Normally
we achieve good results with this setup (up to 2km range outdoors).
Unfortunately it doesn't work well if sender and receiver are close together. In
this case the amplified signal is apparently to strong for the card to handle.
However, switching to the antenna port without the amplifier gives very good
results at close range.
Now we are looking for a way to automatically select the best antenna port. It
seems to me, that this problem is closely related to the rate selection problem
faced by the 802.11 stack, so enhancing one of the rate selection algorithms to
include the antenna into the search space might be the way to go.
Has anybody ever done this? Do you think that this might be the way to go? Are
there better heuristics to find the best antenna port? Any other comments?

Thanks in advance
  Joerg


___
ath5k-devel mailing list
ath5k-devel@lists.ath5k.org
https://lists.ath5k.org/mailman/listinfo/ath5k-devel

Re: [ath5k-devel] Combined rate and antenna selection algorithm

2012-03-15 Thread Adrian Chadd 
Hi!

There's a couple of things I can think of.

Firstly, it'd be nice if we had per-packet TPC working on some of
those earlier chips. Then you could throw in TX power into the mix.

Secondly, yes - you can do per-packet TX antenna selection on these
NICs. You could modify the rate control API in mac80211/ath5k to let
you walk the list of available antennas and do some probing to
determine which antenna gives you the lowest PER or something. The
BSD/madwifi ath(4) driver would do dynamic antenna selection (which
worked on for STA mode, kind of) which would look at the antenna
selected by the hardware on RX and use that as the default.. it
leveraged using fast antenna diversity on the chip and paid attention
to which antenna the hardware would choose. I don't know how much of
that code is left in ath5k; it however is based on RX rather than TX
statistics. It's useful to look at though.


Adrian
___
ath5k-devel mailing list
ath5k-devel@lists.ath5k.org
https://lists.ath5k.org/mailman/listinfo/ath5k-devel