That's why I don't pay attention to efficiency % all that much. It is
fragments received vs fragments expected. As far as packet loss, we've
discussed that here before... that's the beauty of the Canopy scheduler
and retransmission scheme. Retransmit the little 64-byte fragments
instead of the entire frame like wifi. I would imagine at some % of
retransmissions (whole lot of fragments in the frame are missing such
that it cannot be reconstructed and/or not enough time to do so), it
will back the modulation down to minimize retransmissions and the
potential for entire frame loss.
I have seen a very, very nice improvement with 13.2 beta on SM links
that are good in the power level and SNR departments, but the linktests
weren't all that great / what you would expect. Some at say -67 with
25dB SNR would run about 50-55Mbps aggregate and now do 75-80+. I was
seeing very bad multipath on a sector and the SMs would drop session
frequently on 13.1.3. Loaded 13.2 beta and they have not dropped session
even once. I think this is more due to the beacon now being on both
polarities.
On 10/8/2014 2:24 PM, Eric Muehleisen via Af wrote:
I agree with this, but would expect the SM to modulate down from 8X to
6X if the efficiency is low. It seems to be stuck in 8X despite
needing to be in 4X or 6X mod.
BTW. At what point can you expect to see packet loss? It would be nice
to be able to run a linktest and specify a max modulation. What does
72% efficiency at 8X mean? Will efficiency go up if it drops to 6X or
run 100% if set to 4X?
On Wed, Oct 8, 2014 at 2:15 PM, Bill Prince via Af <af@afmug.com
<mailto:af@afmug.com>> wrote:
I think the link efficiency is relative to the current mod mode.
So you can have a high efficiency, even if you're only getting
QPSK modulation.
Conversely, if you're getting poor efficiency but modulation at
256QAM, it can "look" worse.
In my opinion, the efficiency should be relative to the highest
mod mode, regardless of which mod mode you are running in.
So an 8X/1X at 100% efficiency doesn't look better than 50%
efficiency at 8X/8X.
That would make 100% efficiency at 8X/1X look like at least
relative to the total theoretical throughput instead of looking
stellar at a piss-poor modulation mode.
bp
On 10/8/2014 12:03 PM, Eric Muehleisen via Af wrote:
I understand the difference in uplink. My concern is why the SM
with better downlink efficiency delivers much lower throughput.
I've seen this on other AP's as well. Maybe I'm missing something.
On Wed, Oct 8, 2014 at 1:53 PM, George Skorup (Cyber
Broadcasting) via Af <af@afmug.com <mailto:af@afmug.com>> wrote:
Your screenshots appear to be from the SM linktest pages.
Notice one is transmitting 4X, thus the 9Mbps uplink. The
other is 6X and 14Mbps uplink.
For the downlink, look at the AP's session page to see what
it's transmitting to those SMs and you'll have your answer.
On 10/8/2014 1:38 PM, Eric Muehleisen via Af wrote:
https://dl.dropboxusercontent.com/u/1578608/Public/450_linktest.jpg
Two different SM's on the same AP with very similar
signal qualities, QoS, etc. However, the linktests are
vey different. Has anyone see this before? Software 3.1.3.