2009/10/28 Tom DeReggi <wirelessn...@rapiddsl.net>:
> Its relevent to disclose the radio OS type using. (You stated using a R5H a
> Mikrotik card, but weren't clear if using Mikrotik OS).
> The symptom you are explaining sounds similar to how some of my Mikrotik OS
> units had responsed to noise.
> Basically they kept dropping speed until they disconnected. It was like
> watching a clock tick down to zero, and repeat.  I had this problem recently
> with 900Mhz and MIkrotik, and the problem was curred as soon as I switch to
> a different brand product.  I'm suspect, but not verified, nor conclusive,
> that it could be a Mikrotik driver issue.  The point I'm making is that you
> are likely getting some sort of noise or multi-path (self noise), but the
> overall problem may not be the noise/multi-path but instead the inabilty of
> your product to adequately deal with that noise/multipath RF conditions.

Yup, it is Mikrotik 4.1 at both ends, on Routerboard 433AH boards, fed
by a 24v DC plant (batteries and charger).

> The easiest place to do a science project probably isn't between two towers
> seperated by a 20 mile body of water, but it would make for a very
> interesting and meaningful science project.

The link doesn't have traffic over it, the site is currently fed by a
T1, so I have some time to play mad scientist without any negative
effects to customers.

> Sure a Mimo card w/ Dual Pol (for single stream) or Space Diversity would
> likely help deal with Multipath. But what I wonder is whether the same bad
> results are replicated with other single channel products of similar spec.
> It would be interesting to put up a Tlink-45, get results of it's noise
> survey scan, and see if it overcomes the problem.  Or even try a StarOS box.
> Its also relevent to understand how much of the issue is canceled RF, and
> how much is side effects of 802.11 CDMA? I'm wondering if a TDD system w/
> good ARQ better handles it.

I do have a TLink-45 pair sitting around (non connectorized) that I
could test. Is there a hack to put a pigtail on these? If the power is
turned down on the troublesome end, only possible during the times of
day when the RX level is decent, the bandwidth test runs faster and
longer before it drops to nothing. This might make the link usable, if
Mikrotik had some sort of variable transmit power control to maintain
10-20db SNR.

> My point here is in an ideal world a radio should never have the
> characteristic to start at 6mb and slowly go to Zero. Instead it should stay
> at 6mbps, and just have a very high error count. Even if it has 50% packet
> loss it should stay associated, and with a TDD system w/ARQ it likely
> would..

When performing the test, the amount of retransmissions push the data
rate down from 54 to progressively lower modulation speeds. I am
running 5Mhz channels (tried 10, and 20) so this explains the
progressive drop to low throughput and ultimate disconnection in my
mind.

> Also note, if Using Mikrotik, they now support  Atheros's threshold feature,
> to mask out weak signals, this can help reduce multipath signal.
> (although use cutiously as there can be significant fade of water with
> Fog/Clouds/evaporation/Solar and such.)
>
> Obviously if you switched to a TDD MIMO system, you'd optimize your chance
> for success, but you would not be able to learn what factor most helped the
> improvement.


--------------------------------------------------------------------------------
WISPA Wants You! Join today!
http://signup.wispa.org/
--------------------------------------------------------------------------------
 
WISPA Wireless List: wireless@wispa.org

Subscribe/Unsubscribe:
http://lists.wispa.org/mailman/listinfo/wireless

Archives: http://lists.wispa.org/pipermail/wireless/

Reply via email to